[{"data":1,"prerenderedAt":1741},["ShallowReactive",2],{"/blog/there-is-no-mlops-without-devsecops":3,"navigation-en-us":957,"banner-en-us":1385,"footer-en-us":1395,"blog-post-authors-en-us-William Arias":1638,"blog-related-posts-en-us-there-is-no-mlops-without-devsecops":1653,"blog-promotions-en-us":1678,"next-steps-en-us":1731},{"id":4,"title":5,"authors":6,"body":8,"category":936,"date":937,"description":938,"extension":939,"externalUrl":940,"faq":940,"featured":941,"heroImage":942,"meta":943,"navigation":165,"path":944,"seo":945,"slug":949,"stem":950,"tags":951,"template":955,"updatedDate":940,"__hash__":956},"blogPosts/en-us/blog/there-is-no-mlops-without-devsecops.md","Building GitLab with GitLab: Why there is no MLOps without DevSecOps",[7],"William Arias",{"type":9,"value":10,"toc":928},"minimark",[11,15,28,31,41,52,55,60,63,70,73,76,82,85,94,98,101,112,121,127,134,151,490,496,511,517,526,530,533,556,559,562,568,571,619,656,659,696,704,713,719,733,743,749,752,755,759,762,768,771,774,780,783,787,796,802,805,859,865,884,887,900,905,908,924],[12,13,14],"p",{},"Building predictive models requires a good amount of experimentation and iterations. Data scientists building those models usually implement workflows involving several steps such as data loading, processing, training, testing, and deployment. Such workflows or data science pipelines come with a set of challenges on their own; some of these common challenges are:",[16,17,18,22,25],"ul",{},[19,20,21],"li",{},"prone to error due to manual steps",[19,23,24],{},"experimentation results that are hard to replicate",[19,26,27],{},"long training time of machine learning (ML) models",[12,29,30],{},"When there is a challenge, there is also an opportunity; in this case, those challenges represent an opportunity for data scientists to adopt DevSecOps practices and enjoy the benefits of automation, repeatable workflows, standardization, and automatic provisioning of infrastructure needed for data-driven applications at scale.",[12,32,33,34,40],{},"The ",[35,36,39],"a",{"href":37,"rel":38},"https://handbook.gitlab.com/handbook/enterprise-data/organization/data-science/",[],"Data Science team at\nGitLab","\nis now utilizing the GitLab DevSecOps Platform in their workflows, specifically to:",[16,42,43,46,49],{},[19,44,45],{},"enhance experiment reproducibility by ensuring code and data execute in a\nstandardized container image",[19,47,48],{},"automate training and re-training of ML models with GPU-enabled CI/CD",[19,50,51],{},"leverage ML experiment tracking, storing the most relevant metadata and\nartifacts produced by data science pipelines automated with CI",[12,53,54],{},"At GitLab, we are proponents of \"dogfooding\" our platform and sharing how we use GitLab to build GitLab. What follows is a detailed look at the Data\nScience team's experience.",[56,57,59],"h3",{"id":58},"enhancing-experiment-reproducibility","Enhancing experiment reproducibility",[12,61,62],{},"A baseline step to enhance reproducibility is having a common and standard experiment environment for all data scientists to run experiments in their\nJupyter Notebooks. A standard data science environment ensures that all team members use the same software dependencies. A way to achieve this is by building a container image with all the respective dependencies under version control and re-pulling it every time a new version of the code is run. This process is illustrated in the figure below:",[12,64,65],{},[66,67],"img",{"alt":68,"src":69},"build","https://res.cloudinary.com/about-gitlab-com/image/upload/v1782397850/blog/Content%20Images/2023-10-04-there-is-no-mlops-without-devsecops/build-2.png",[12,71,72],{},"Data science image of automatic build using GitLab CI",[12,74,75],{},"You might wonder if the image gets built every time there is a new commit.\nThe answer is \"no\" since that would result in longer execution times, and the image dependencies versions don’t change frequently, rendering it unnecessary to build it every time there is a new commit. Therefore, once the standard image is automatically built by the pipeline, it is pushed to the GitLab Container Registry, where it is stored and ready to be pulled every time changes to the model code are introduced, and re-training is necessary.",[12,77,78],{},[66,79],{"alt":80,"src":81},"registry","https://res.cloudinary.com/about-gitlab-com/image/upload/v1782397851/blog/Content%20Images/2023-10-04-there-is-no-mlops-without-devsecops/registry.png",[12,83,84],{},"GitLab Container Registry with image automatically built and pushed by a CI pipeline",[12,86,87,88,93],{},"Changes to the image dependencies or Dockerfile require a ",[35,89,92],{"href":90,"rel":91},"https://docs.gitlab.com/user/project/merge_requests/",[],"merge request"," and an approval process.",[56,95,97],{"id":96},"how-to-build-the-data-science-image-using-gitlab-cicd","How to build the data science image using GitLab CI/CD",[12,99,100],{},"Consider this project structure:",[102,103,109],"pre",{"className":104,"code":106,"language":107,"meta":108},[105],"language-text","notebooks/\n.gitlab-ci.yml\nDockerfile\nconfig.yml\nrequirements.txt\n","text","",[110,111,106],"code",{"__ignoreMap":108},[12,113,114,115,120],{},"GitLab's Data Science team already had a pre-configured JupyterLab image with packages such as ",[35,116,119],{"href":117,"rel":118},"https://pypi.org/project/gitlabds/1.0.0/",[],"gitlabds","\nfor common data preparation tasks and modules to enable Snowflake connectivity for loading raw data. All these dependencies are reflected in the Dockerfile at the root of the project, plus all the steps necessary to build the image:",[102,122,125],{"className":123,"code":124,"language":107,"meta":108},[105],"FROM nvcr.io/nvidia/cuda:12.1.1-base-ubuntu22.04\nCOPY .    /app/\nWORKDIR /app\nRUN apt-get update\nRUN apt-get install -y python3.9\nRUN apt-get install -y python3-pip\nRUN pip install -r requirements.txt\n",[110,126,124],{"__ignoreMap":108},[12,128,129,130,133],{},"The instructions to build the data science image start with using Ubuntu with CUDA drivers as a base image. We are using this baseline image because, moving forward, we will use GPU hardware to train models. The rest of the steps include installing Python 3.9 and the dependencies listed in ",[110,131,132],{},"requirements.txt"," with their respective versions.",[12,135,136,137,142,143,146,147,150],{},"Automatically building the data science image using ",[35,138,141],{"href":139,"rel":140},"https://about.gitlab.com/topics/ci-cd/",[],"GitLab\nCI/CD"," requires us to create the ",[110,144,145],{},".gitlab-ci.yml "," at the root of the project and use it to describe the jobs we want to automate. For the time being, let’s focus only on the ",[110,148,149],{},"build-ds-image","job:",[102,152,156],{"className":153,"code":154,"language":155,"meta":108,"style":108},"language-yaml shiki shiki-themes github-light","\nvariables:\n  DOCKER_HOST: tcp://docker:2375\n  MOUNT_POINT: \"/builds/$CI_PROJECT_PATH/mnt\"\n  CONTAINER_IMAGE: \"$CI_REGISTRY_IMAGE/main-image:latest\"\n\nstages:\n    - build\n    - train\n    - notify\ninclude:\n  - template: 'Workflows/MergeRequest-Pipelines.gitlab-ci.yml'\nworkflow:\n  rules:\n    - if: $CI_PIPELINE_SOURCE == \"merge_request_event\"\n    - if: $CI_COMMIT_BRANCH && $CI_OPEN_MERGE_REQUESTS\n      when: never\n\nbuild-ds-image:\n  tags: [ saas-linux-large-amd64 ]\n  stage: build\n  services:\n    - docker:20.10.16-dind\n  image:\n    name: docker:20.10.16\n  script:\n    - docker login -u $CI_REGISTRY_USER -p $CI_REGISTRY_PASSWORD $CI_REGISTRY\n    - docker build -t $CONTAINER_IMAGE .\n    - docker push $CONTAINER_IMAGE\n  rules:\n    - if: '$CI_PIPELINE_SOURCE == \"merge_request_event\" && $CI_MERGE_REQUEST_TARGET_BRANCH_NAME == $CI_DEFAULT_BRANCH'\n      changes:\n        - Dockerfile\n        - requirements.txt\n\n  allow_failure: true\n","yaml",[110,157,158,167,178,191,202,213,218,226,235,243,251,259,273,281,289,302,314,325,330,337,352,362,370,378,386,397,405,413,421,429,436,448,456,465,473,478],{"__ignoreMap":108},[159,160,163],"span",{"class":161,"line":162},"line",1,[159,164,166],{"emptyLinePlaceholder":165},true,"\n",[159,168,170,174],{"class":161,"line":169},2,[159,171,173],{"class":172},"shJU0","variables",[159,175,177],{"class":176},"sgsFI",":\n",[159,179,181,184,187],{"class":161,"line":180},3,[159,182,183],{"class":172},"  DOCKER_HOST",[159,185,186],{"class":176},": ",[159,188,190],{"class":189},"sYBdl","tcp://docker:2375\n",[159,192,194,197,199],{"class":161,"line":193},4,[159,195,196],{"class":172},"  MOUNT_POINT",[159,198,186],{"class":176},[159,200,201],{"class":189},"\"/builds/$CI_PROJECT_PATH/mnt\"\n",[159,203,205,208,210],{"class":161,"line":204},5,[159,206,207],{"class":172},"  CONTAINER_IMAGE",[159,209,186],{"class":176},[159,211,212],{"class":189},"\"$CI_REGISTRY_IMAGE/main-image:latest\"\n",[159,214,216],{"class":161,"line":215},6,[159,217,166],{"emptyLinePlaceholder":165},[159,219,221,224],{"class":161,"line":220},7,[159,222,223],{"class":172},"stages",[159,225,177],{"class":176},[159,227,229,232],{"class":161,"line":228},8,[159,230,231],{"class":176},"    - ",[159,233,234],{"class":189},"build\n",[159,236,238,240],{"class":161,"line":237},9,[159,239,231],{"class":176},[159,241,242],{"class":189},"train\n",[159,244,246,248],{"class":161,"line":245},10,[159,247,231],{"class":176},[159,249,250],{"class":189},"notify\n",[159,252,254,257],{"class":161,"line":253},11,[159,255,256],{"class":172},"include",[159,258,177],{"class":176},[159,260,262,265,268,270],{"class":161,"line":261},12,[159,263,264],{"class":176},"  - ",[159,266,267],{"class":172},"template",[159,269,186],{"class":176},[159,271,272],{"class":189},"'Workflows/MergeRequest-Pipelines.gitlab-ci.yml'\n",[159,274,276,279],{"class":161,"line":275},13,[159,277,278],{"class":172},"workflow",[159,280,177],{"class":176},[159,282,284,287],{"class":161,"line":283},14,[159,285,286],{"class":172},"  rules",[159,288,177],{"class":176},[159,290,292,294,297,299],{"class":161,"line":291},15,[159,293,231],{"class":176},[159,295,296],{"class":172},"if",[159,298,186],{"class":176},[159,300,301],{"class":189},"$CI_PIPELINE_SOURCE == \"merge_request_event\"\n",[159,303,305,307,309,311],{"class":161,"line":304},16,[159,306,231],{"class":176},[159,308,296],{"class":172},[159,310,186],{"class":176},[159,312,313],{"class":189},"$CI_COMMIT_BRANCH && $CI_OPEN_MERGE_REQUESTS\n",[159,315,317,320,322],{"class":161,"line":316},17,[159,318,319],{"class":172},"      when",[159,321,186],{"class":176},[159,323,324],{"class":189},"never\n",[159,326,328],{"class":161,"line":327},18,[159,329,166],{"emptyLinePlaceholder":165},[159,331,333,335],{"class":161,"line":332},19,[159,334,149],{"class":172},[159,336,177],{"class":176},[159,338,340,343,346,349],{"class":161,"line":339},20,[159,341,342],{"class":172},"  tags",[159,344,345],{"class":176},": [ ",[159,347,348],{"class":189},"saas-linux-large-amd64",[159,350,351],{"class":176}," ]\n",[159,353,355,358,360],{"class":161,"line":354},21,[159,356,357],{"class":172},"  stage",[159,359,186],{"class":176},[159,361,234],{"class":189},[159,363,365,368],{"class":161,"line":364},22,[159,366,367],{"class":172},"  services",[159,369,177],{"class":176},[159,371,373,375],{"class":161,"line":372},23,[159,374,231],{"class":176},[159,376,377],{"class":189},"docker:20.10.16-dind\n",[159,379,381,384],{"class":161,"line":380},24,[159,382,383],{"class":172},"  image",[159,385,177],{"class":176},[159,387,389,392,394],{"class":161,"line":388},25,[159,390,391],{"class":172},"    name",[159,393,186],{"class":176},[159,395,396],{"class":189},"docker:20.10.16\n",[159,398,400,403],{"class":161,"line":399},26,[159,401,402],{"class":172},"  script",[159,404,177],{"class":176},[159,406,408,410],{"class":161,"line":407},27,[159,409,231],{"class":176},[159,411,412],{"class":189},"docker login -u $CI_REGISTRY_USER -p $CI_REGISTRY_PASSWORD $CI_REGISTRY\n",[159,414,416,418],{"class":161,"line":415},28,[159,417,231],{"class":176},[159,419,420],{"class":189},"docker build -t $CONTAINER_IMAGE .\n",[159,422,424,426],{"class":161,"line":423},29,[159,425,231],{"class":176},[159,427,428],{"class":189},"docker push $CONTAINER_IMAGE\n",[159,430,432,434],{"class":161,"line":431},30,[159,433,286],{"class":172},[159,435,177],{"class":176},[159,437,439,441,443,445],{"class":161,"line":438},31,[159,440,231],{"class":176},[159,442,296],{"class":172},[159,444,186],{"class":176},[159,446,447],{"class":189},"'$CI_PIPELINE_SOURCE == \"merge_request_event\" && $CI_MERGE_REQUEST_TARGET_BRANCH_NAME == $CI_DEFAULT_BRANCH'\n",[159,449,451,454],{"class":161,"line":450},32,[159,452,453],{"class":172},"      changes",[159,455,177],{"class":176},[159,457,459,462],{"class":161,"line":458},33,[159,460,461],{"class":176},"        - ",[159,463,464],{"class":189},"Dockerfile\n",[159,466,468,470],{"class":161,"line":467},34,[159,469,461],{"class":176},[159,471,472],{"class":189},"requirements.txt\n",[159,474,476],{"class":161,"line":475},35,[159,477,166],{"emptyLinePlaceholder":165},[159,479,481,484,486],{"class":161,"line":480},36,[159,482,483],{"class":172},"  allow_failure",[159,485,186],{"class":176},[159,487,489],{"class":488},"sYu0t","true\n",[12,491,492,493,495],{},"At a high level, the job ",[110,494,149],{},":",[16,497,498,501,504],{},[19,499,500],{},"uses a docker-in-docker service (dind) necessary to create docker images\nin GitLab CI/CD.",[19,502,503],{},"uses predefined variables to log into the GitLab Container\nRegistry, build the image, tag it using $CONTAINER_IMAGE variable, and push it to the registry. These steps are declared in the script section lines.",[19,505,506,507,510],{},"leverages a  ",[110,508,509],{},"rules"," section to evaluate conditions to determine if the\njob should be created. In this case, this job runs only if there are changes to the Dockerfile and requirements.txt file and if those changes are created using a merge request.",[12,512,513,514,516],{},"The conditions declared in ",[110,515,509],{}," helps us optimize the pipeline running time since the image gets rebuilt only when necessary.",[12,518,519,520,525],{},"A complete pipeline can be found in this example project, along with instructions to trigger the automatic creation of the data science image:\n",[35,521,524],{"href":522,"rel":523},"https://gitlab.com/gitlab-data/data-science-ci-example/-/blob/main/.gitlab-ci.yml?ref_type=heads",[],"Data Science CI pipeline",".",[56,527,529],{"id":528},"automate-training-and-re-training-of-ml-models-with-gpu-enabled-cicd","Automate training and re-training of ML models with GPU-enabled CI/CD",[12,531,532],{},"GitLab offers the ability to leverage GPU hardware and, even better, to get this hardware automatically provisioned to run jobs declared in the\n.gitlab-ci.yml file. We took advantage of this capability to train our ML models faster without spending time setting up or configuring graphics card drivers. Using GPU hardware (GitLab\nRunners)\nrequires us to add this line to the training job:",[102,534,536],{"className":153,"code":535,"language":155,"meta":108,"style":108},"\ntags:\n        - saas-linux-medium-amd64-gpu-standard\n",[110,537,538,542,549],{"__ignoreMap":108},[159,539,540],{"class":161,"line":162},[159,541,166],{"emptyLinePlaceholder":165},[159,543,544,547],{"class":161,"line":169},[159,545,546],{"class":172},"tags",[159,548,177],{"class":176},[159,550,551,553],{"class":161,"line":180},[159,552,461],{"class":176},[159,554,555],{"class":189},"saas-linux-medium-amd64-gpu-standard\n",[12,557,558],{},"The tag above will ensure that a GPU GitLab Runner automatically picks up every training job.",[12,560,561],{},"Let’s take a look at the entire training job in the .gitlab-ci.yml file and break down what it does:",[102,563,566],{"className":564,"code":565,"language":107,"meta":108},[105],"\ntrain-commit-activated:\n    stage: train\n    image: $CONTAINER_IMAGE\n    tags:\n        - saas-linux-medium-amd64-gpu-standard\n    script:\n        - echo \"GPU training activated by commit message\"\n        - echo \"message passed is $CI_COMMIT_MESSAGE\"\n        - notebookName=$(echo ${CI_COMMIT_MESSAGE/train})\n        - echo \"Notebook name $notebookName\"\n        - papermill -p is_local_development False -p tree_method 'gpu_hist' $notebookName -\n    rules:\n        - if: '$CI_COMMIT_BRANCH == \"staging\"'\n          when: never\n        - if: $CI_COMMIT_MESSAGE =~ /\\w+\\.ipynb/\n          when: always\n          allow_failure: true\n    artifacts:\n      paths:\n        - ./model_metrics.md\n",[110,567,565],{"__ignoreMap":108},[12,569,570],{},"Let’s start with this block:",[102,572,574],{"className":153,"code":573,"language":155,"meta":108,"style":108},"\ntrain-commit-activated:\n    stage: train\n    image: $CONTAINER_IMAGE\n    tags:\n        - saas-linux-medium-amd64-gpu-standard\n",[110,575,576,580,587,596,606,613],{"__ignoreMap":108},[159,577,578],{"class":161,"line":162},[159,579,166],{"emptyLinePlaceholder":165},[159,581,582,585],{"class":161,"line":169},[159,583,584],{"class":172},"train-commit-activated",[159,586,177],{"class":176},[159,588,589,592,594],{"class":161,"line":180},[159,590,591],{"class":172},"    stage",[159,593,186],{"class":176},[159,595,242],{"class":189},[159,597,598,601,603],{"class":161,"line":193},[159,599,600],{"class":172},"    image",[159,602,186],{"class":176},[159,604,605],{"class":189},"$CONTAINER_IMAGE\n",[159,607,608,611],{"class":161,"line":204},[159,609,610],{"class":172},"    tags",[159,612,177],{"class":176},[159,614,615,617],{"class":161,"line":215},[159,616,461],{"class":176},[159,618,555],{"class":189},[16,620,621,627,644,650],{},[19,622,623,626],{},[624,625,584],"strong",{}," This is the name of the job. Since the model\ntraining gets activated given a specific pattern in the commit message, we use a descriptive name to easily identify it in the larger pipeline.",[19,628,629,632,633,635,636,639,640,643],{},[624,630,631],{},"stage: train"," This specifies the pipeline stage where this job belongs.\nIn the first part of the CI/CD configuration, we defined three stages for this pipeline: ",[110,634,68],{},", ",[110,637,638],{},"train",",  and ",[110,641,642],{},"notify",". This job comes after building the data science container image. The order is essential since we first need the image built to run our training code in it.",[19,645,646,649],{},[624,647,648],{},"image: $CONTAINER_IMAGE"," Here, we specify the Docker image built in the\nfirst job that contains the CUDA drivers and necessary Python dependencies to run this job. $CONTAINER_IMAGE is a user-defined variable specified in the variables section of the .gitlab-ci.yml file.",[19,651,652,655],{},[624,653,654],{},"tags: saas-linux-medium-amd64-gpu-standard"," As mentioned earlier, using\nthis line, we ask GitLab to automatically provision a GPU-enabled Runner to execute this job.",[12,657,658],{},"The second block of the job:",[102,660,664],{"className":661,"code":662,"language":663,"meta":108,"style":108},"language-markdown shiki shiki-themes github-light","script:\n        - echo \"GPU training activated by commit message\"\n        - echo \"message passed is $CI_COMMIT_MESSAGE\"\n        - notebookName=$(echo ${CI_COMMIT_MESSAGE/train})\n        - echo \"Notebook name $notebookName\"\n        - papermill -p is_local_development False -p tree_method 'gpu_hist' $notebookName -\n","markdown",[110,665,666,671,676,681,686,691],{"__ignoreMap":108},[159,667,668],{"class":161,"line":162},[159,669,670],{"class":176},"script:\n",[159,672,673],{"class":161,"line":169},[159,674,675],{"class":176},"        - echo \"GPU training activated by commit message\"\n",[159,677,678],{"class":161,"line":180},[159,679,680],{"class":176},"        - echo \"message passed is $CI_COMMIT_MESSAGE\"\n",[159,682,683],{"class":161,"line":193},[159,684,685],{"class":176},"        - notebookName=$(echo ${CI_COMMIT_MESSAGE/train})\n",[159,687,688],{"class":161,"line":204},[159,689,690],{"class":176},"        - echo \"Notebook name $notebookName\"\n",[159,692,693],{"class":161,"line":215},[159,694,695],{"class":176},"        - papermill -p is_local_development False -p tree_method 'gpu_hist' $notebookName -\n",[16,697,698],{},[19,699,700,703],{},[624,701,702],{},"script"," This section contains the commands in charge of running the\nmodel training. The execution of this job is conditioned to the contents of the  commit message. The commit message must have the name of the Jupyter\nNotebook that contains the actual model training code.",[12,705,706,707,712],{},"The rationale behind this approach is that we wanted to keep the data scientist workflow as simple as possible. The team had already adopted the ",[35,708,711],{"href":709,"rel":710},"https://gitlab.com/gitlab-data/data-science/-/tree/main/templates",[],"modeling templates","\nto start building predictive models quickly. Plugging the CI pipeline into their modeling workflow was a priority to ensure productivity would remain intact. With these steps:",[102,714,717],{"className":715,"code":716,"language":107,"meta":108},[105],"notebookName=$(echo ${CI_COMMIT_MESSAGE/train})\n        - echo \"Notebook name $notebookName\"\n        - papermill -p is_local_development False -p tree_method 'gpu_hist' $notebookName -\n",[110,718,716],{"__ignoreMap":108},[12,720,721,722,727,728,525],{},"The CI pipeline captures the name of the Jupyter Notebook with the training modeling template and passes parameters to ensure ",[35,723,726],{"href":724,"rel":725},"https://xgboost.readthedocs.io/en/stable/",[],"XGBoost"," uses the provisioned\nGPU. You can find an example of the Jupyter modeling template that is executed in this job ",[35,729,732],{"href":730,"rel":731},"https://gitlab.com/gitlab-data/data-science-ci-example/-/blob/main/notebooks/training_example.ipynb?ref_type=heads",[],"here",[12,734,735,736,738,739,742],{},"Once the data science image is built, it can be reutilized in further model training jobs. The ",[110,737,584],{}," job pulls the image from the\nGitLab Container Registry and utilizes it to run the ML pipeline defined in the training notebook. This is illustrated in the ",[110,740,741],{},"CI Job - Train model"," in the figure below:",[12,744,745],{},[66,746],{"alt":747,"src":748},"training","https://res.cloudinary.com/about-gitlab-com/image/upload/v1782397851/blog/Content%20Images/2023-10-04-there-is-no-mlops-without-devsecops/training_job.png",[12,750,751],{},"Training job executes ML pipeline defined in the modeling notebook",[12,753,754],{},"Since our image contains CUDA drivers and GitLab automatically provisions\nGPU-enabled hardware, the training job runs significantly faster with respect to standard hardware.",[56,756,758],{"id":757},"using-gitlab-ml-experiment-tracker","Using GitLab ML experiment tracker",[12,760,761],{},"Each model training execution triggered using GitLab CI is an experiment that needs tracking. Using Experiment tracking in GitLab helps us to record metadata that comes in handy to compare model performance and collaborate with other data scientists by making result experiments available for everyone and providing a detailed history of the model development.",[12,763,764],{},[66,765],{"alt":766,"src":767},"experiments","https://res.cloudinary.com/about-gitlab-com/image/upload/v1782397850/blog/Content%20Images/2023-10-04-there-is-no-mlops-without-devsecops/experiments.png",[12,769,770],{},"Experiments automatically logged on every CI pipeline GPU training run",[12,772,773],{},"Each model artifact created can be traced back to the pipeline that generated it, along with its dependencies:",[12,775,776],{},[66,777],{"alt":778,"src":779},"traceability","https://res.cloudinary.com/about-gitlab-com/image/upload/v1782397852/blog/Content%20Images/2023-10-04-there-is-no-mlops-without-devsecops/traceability_small.png",[12,781,782],{},"Model traceability from pipeline run to candidate details",[56,784,786],{"id":785},"putting-it-all-together","Putting it all together",[12,788,789,790,795],{},"What is machine learning without data to learn from? We also leveraged the ",[35,791,794],{"href":792,"rel":793},"https://www.snowflake.com/en/",[],"Snowflake"," connector in the model training notebook and automated the data extraction whenever the respective commit triggers a training job. Here is an architecture of the current solution with all the parts described in this blog post:",[12,797,798],{},[66,799],{"alt":800,"src":801},"process","https://res.cloudinary.com/about-gitlab-com/image/upload/v1782397851/blog/Content%20Images/2023-10-04-there-is-no-mlops-without-devsecops/training_fixed.png",[12,803,804],{},"Data Science pipelines automated using GitLab DevSecops Platform",[806,807,808,821],"table",{},[809,810,811],"thead",{},[812,813,814,818],"tr",{},[815,816,817],"th",{},"Challenge",[815,819,820],{},"Solution",[822,823,824,838,851],"tbody",{},[812,825,826,830],{},[827,828,829],"td",{},"Prone to error due to manual steps",[827,831,832,833],{},"Automate steps with ",[35,834,837],{"href":835,"rel":836},"https://docs.gitlab.com/ci/",[],"GitLab CI/CD",[812,839,840,843],{},[827,841,842],{},"Experimentation results that are hard to replicate",[827,844,845,846],{},"Record metadata and model artifacts with ",[35,847,850],{"href":848,"rel":849},"https://docs.gitlab.com/user/project/ml/experiment_tracking/",[],"GitLab Experiment Tracker",[812,852,853,856],{},[827,854,855],{},"The long training time of machine learning models",[827,857,858],{},"Train models with GitLab SaaS GPU Runners",[12,860,861,862,525],{},"Iterating on these challenges is a first step towards MLOps, and we are at the tip of the iceberg; in coming iterations, we will adopt security features to ensure model provenance (software bill of materials) and code quality, and to monitor our ML workflow development with value stream dashboards. But so far, one thing is sure: ",[624,863,864],{},"There is no MLOps without\nDevSecOps",[12,866,867,868,873,874,879,880,883],{},"Get started automating your data science pipelines, follow this ",[35,869,872],{"href":870,"rel":871},"https://handbook.gitlab.com/handbook/enterprise-data/platform/ci-for-ds-pipelines/",[],"tutorial","\nand clone this ",[35,875,878],{"href":876,"rel":877},"https://gitlab.com/gitlab-data/data-science-ci-example",[],"data-science-project","\nto follow along and watch this demo of using GPU Runners to train ",[35,881,726],{"href":724,"rel":882},[]," model.",[12,885,886],{},"See how data scientists can train ML models with GitLab GPU-enabled Runners (XGBoost 5-minute demo):",[888,889,892,893],"figure",{"className":890},[891],"video_container","\n  ",[894,895,899],"iframe",{"src":896,"frameBorder":897,"allowFullScreen":898},"https://www.youtube.com/embed/tElegG4NCZ0?si=L1IZfx_UGv6u81Gk","0","true"," ",[901,902,904],"h2",{"id":903},"more-building-gitlab-with-gitlab-blogs","More \"Building GitLab with GitLab\" blogs",[12,906,907],{},"Read more of our \"Building GitLab with GitLab\" series:",[16,909,910,917],{},[19,911,912],{},[35,913,916],{"href":914,"rel":915},"https://about.gitlab.com/blog/building-gitlab-with-gitlab-api-fuzzing-workflow/",[],"How we use Web API fuzz\ntesting",[19,918,919],{},[35,920,923],{"href":921,"rel":922},"https://about.gitlab.com/blog/building-gitlab-with-gitlabcom-how-gitlab-inspired-dedicated/",[],"How GitLab.com inspired GitLab\nDedicated",[925,926,927],"style",{},"html pre.shiki code .shJU0, html code.shiki .shJU0{--shiki-default:#22863A}html pre.shiki code .sgsFI, html code.shiki .sgsFI{--shiki-default:#24292E}html pre.shiki code .sYBdl, html code.shiki .sYBdl{--shiki-default:#032F62}html pre.shiki code .sYu0t, html code.shiki .sYu0t{--shiki-default:#005CC5}html .default .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}",{"title":108,"searchDepth":169,"depth":169,"links":929},[930,931,932,933,934,935],{"id":58,"depth":180,"text":59},{"id":96,"depth":180,"text":97},{"id":528,"depth":180,"text":529},{"id":757,"depth":180,"text":758},{"id":785,"depth":180,"text":786},{"id":903,"depth":169,"text":904},"ai","2023-10-05","Follow along as data scientists adopt DevSecOps practices and enjoy the benefits of automation, repeatable workflows, standardization, and automatic provisioning of infrastructure.","md",null,false,"https://res.cloudinary.com/about-gitlab-com/image/upload/v1749659740/Blog/Hero%20Images/building-gitlab-with-gitlab-no-type.png",{},"/en-us/blog/there-is-no-mlops-without-devsecops",{"ogTitle":5,"ogImage":942,"ogDescription":938,"ogSiteName":946,"noIndex":941,"ogType":947,"ogUrl":948,"title":5,"canonicalUrls":948,"description":938},"https://about.gitlab.com","article","https://about.gitlab.com/blog/there-is-no-mlops-without-devsecops","there-is-no-mlops-without-devsecops","en-us/blog/there-is-no-mlops-without-devsecops",[872,952,953,954],"DevSecOps","DevSecOps platform","AI","BlogPost","oRuLur80bMNz1_aptMw6wUUXGcrcbp8qPBEk-80xfWA",{"logo":958,"freeTrial":963,"sales":968,"login":973,"items":978,"search":1305,"minimal":1336,"duo":1355,"switchNav":1364,"pricingDeployment":1375},{"config":959},{"href":960,"dataGaName":961,"dataGaLocation":962},"/","gitlab logo","header",{"text":964,"config":965},"Get free trial",{"href":966,"dataGaName":967,"dataGaLocation":962},"https://gitlab.com/-/trial_registrations/new?glm_source=about.gitlab.com&glm_content=default-saas-trial/","free trial",{"text":969,"config":970},"Request a demo",{"href":971,"dataGaName":972,"dataGaLocation":962},"/sales/?contact-topic=request-demo","sales",{"text":974,"config":975},"Sign in",{"href":976,"dataGaName":977,"dataGaLocation":962},"https://gitlab.com/users/sign_in/","sign in",[979,1008,1108,1113,1227,1283],{"text":980,"config":981,"menu":983},"Platform",{"dataNavLevelOne":982},"platform",{"type":984,"columns":985},"cards",[986,992,1000],{"title":980,"description":987,"link":988},"The intelligent orchestration platform for DevSecOps",{"text":989,"config":990},"Explore our Platform",{"href":991,"dataGaName":982,"dataGaLocation":962},"/platform/",{"title":993,"description":994,"link":995},"GitLab Duo Agent Platform","Agentic AI for the entire software lifecycle",{"text":996,"config":997},"Meet GitLab Duo",{"href":998,"dataGaName":999,"dataGaLocation":962},"/gitlab-duo-agent-platform/","gitlab duo agent platform",{"title":1001,"description":1002,"link":1003},"Why GitLab","See the top reasons enterprises choose GitLab",{"text":1004,"config":1005},"Learn more",{"href":1006,"dataGaName":1007,"dataGaLocation":962},"/why-gitlab/","why gitlab",{"text":1009,"left":165,"config":1010,"menu":1012},"Product",{"dataNavLevelOne":1011},"solutions",{"type":1013,"link":1014,"columns":1018,"feature":1087},"lists",{"text":1015,"config":1016},"View all Solutions",{"href":1017,"dataGaName":1011,"dataGaLocation":962},"/solutions/",[1019,1043,1066],{"title":1020,"description":1021,"link":1022,"items":1027},"Automation","CI/CD and automation to accelerate deployment",{"config":1023},{"icon":1024,"href":1025,"dataGaName":1026,"dataGaLocation":962},"AutomatedCodeAlt","/solutions/delivery-automation/","automated software delivery",[1028,1032,1035,1039],{"text":1029,"config":1030},"CI/CD",{"href":1031,"dataGaLocation":962,"dataGaName":1029},"/solutions/continuous-integration/",{"text":993,"config":1033},{"href":998,"dataGaLocation":962,"dataGaName":1034},"gitlab duo agent platform - product menu",{"text":1036,"config":1037},"Source Code Management",{"href":1038,"dataGaLocation":962,"dataGaName":1036},"/solutions/source-code-management/",{"text":1040,"config":1041},"Automated Software Delivery",{"href":1025,"dataGaLocation":962,"dataGaName":1042},"Automated software delivery",{"title":1044,"description":1045,"link":1046,"items":1051},"Security","Deliver code faster without compromising security",{"config":1047},{"href":1048,"dataGaName":1049,"dataGaLocation":962,"icon":1050},"/solutions/application-security-testing/","security and compliance","ShieldCheckLight",[1052,1056,1061],{"text":1053,"config":1054},"Application Security Testing",{"href":1048,"dataGaName":1055,"dataGaLocation":962},"Application security testing",{"text":1057,"config":1058},"Software Supply Chain Security",{"href":1059,"dataGaLocation":962,"dataGaName":1060},"/solutions/supply-chain/","Software supply chain security",{"text":1062,"config":1063},"Software Compliance",{"href":1064,"dataGaName":1065,"dataGaLocation":962},"/solutions/software-compliance/","software compliance",{"title":1067,"link":1068,"items":1073},"Measurement",{"config":1069},{"icon":1070,"href":1071,"dataGaName":1072,"dataGaLocation":962},"DigitalTransformation","/solutions/visibility-measurement/","visibility and measurement",[1074,1078,1082],{"text":1075,"config":1076},"Visibility & Measurement",{"href":1071,"dataGaLocation":962,"dataGaName":1077},"Visibility and Measurement",{"text":1079,"config":1080},"Value Stream Management",{"href":1081,"dataGaLocation":962,"dataGaName":1079},"/solutions/value-stream-management/",{"text":1083,"config":1084},"Analytics & Insights",{"href":1085,"dataGaLocation":962,"dataGaName":1086},"/solutions/analytics-and-insights/","Analytics and insights",{"title":1088,"type":1013,"items":1089},"GitLab for",[1090,1096,1102],{"text":1091,"config":1092},"Enterprise",{"icon":1093,"href":1094,"dataGaLocation":962,"dataGaName":1095},"Building","/enterprise/","enterprise",{"text":1097,"config":1098},"Small Business",{"icon":1099,"href":1100,"dataGaLocation":962,"dataGaName":1101},"Work","/small-business/","small business",{"text":1103,"config":1104},"Public Sector",{"icon":1105,"href":1106,"dataGaLocation":962,"dataGaName":1107},"Organization","/solutions/public-sector/","public sector",{"text":1109,"config":1110},"Pricing",{"href":1111,"dataGaName":1112,"dataGaLocation":962,"dataNavLevelOne":1112},"/pricing/","pricing",{"text":1114,"config":1115,"menu":1117},"Resources",{"dataNavLevelOne":1116},"resources",{"type":1013,"link":1118,"columns":1122,"feature":1216},{"text":1119,"config":1120},"View all resources",{"href":1121,"dataGaName":1116,"dataGaLocation":962},"/resources/",[1123,1156,1183],{"title":1124,"items":1125},"Getting started",[1126,1131,1136,1141,1146,1151],{"text":1127,"config":1128},"Install",{"href":1129,"dataGaName":1130,"dataGaLocation":962},"/install/","install",{"text":1132,"config":1133},"Quick start guides",{"href":1134,"dataGaName":1135,"dataGaLocation":962},"/get-started/","quick setup checklists",{"text":1137,"config":1138},"Learn",{"href":1139,"dataGaLocation":962,"dataGaName":1140},"https://university.gitlab.com/","learn",{"text":1142,"config":1143},"Product documentation",{"href":1144,"dataGaName":1145,"dataGaLocation":962},"https://docs.gitlab.com/","product documentation",{"text":1147,"config":1148},"Best practice videos",{"href":1149,"dataGaName":1150,"dataGaLocation":962},"/getting-started-videos/","best practice videos",{"text":1152,"config":1153},"Integrations",{"href":1154,"dataGaName":1155,"dataGaLocation":962},"/integrations/","integrations",{"title":1157,"items":1158},"Discover",[1159,1164,1169,1174,1178],{"text":1160,"config":1161},"Customer success stories",{"href":1162,"dataGaName":1163,"dataGaLocation":962},"/customers/","customer success stories",{"text":1165,"config":1166},"Blog",{"href":1167,"dataGaName":1168,"dataGaLocation":962},"/blog/","blog",{"text":1170,"config":1171},"Demo Hub",{"href":1172,"dataGaName":1173,"dataGaLocation":962},"/demo-hub/","demo hub",{"text":1175,"config":1176},"The Source",{"href":1177,"dataGaName":1168,"dataGaLocation":962},"/the-source/",{"text":1179,"config":1180},"Remote",{"href":1181,"dataGaName":1182,"dataGaLocation":962},"https://handbook.gitlab.com/handbook/company/culture/all-remote/","remote",{"title":1184,"items":1185},"Connect",[1186,1191,1196,1201,1206,1211],{"text":1187,"config":1188},"GitLab Services",{"href":1189,"dataGaName":1190,"dataGaLocation":962},"/services/","services",{"text":1192,"config":1193},"Contribute",{"href":1194,"dataGaName":1195,"dataGaLocation":962},"https://contributors.gitlab.com","contribute",{"text":1197,"config":1198},"Community",{"href":1199,"dataGaName":1200,"dataGaLocation":962},"/community/","community",{"text":1202,"config":1203},"Forum",{"href":1204,"dataGaName":1205,"dataGaLocation":962},"https://forum.gitlab.com/","forum",{"text":1207,"config":1208},"Events",{"href":1209,"dataGaName":1210,"dataGaLocation":962},"/events/","events",{"text":1212,"config":1213},"Partners",{"href":1214,"dataGaName":1215,"dataGaLocation":962},"/partners/","partners",{"config":1217,"title":1220,"text":1221,"link":1222},{"background":1218,"textColor":1219},"url('https://res.cloudinary.com/about-gitlab-com/image/upload/v1777322348/qpq8yrgn8knii57omj0c.png')","#000","What’s new in GitLab","Stay updated with our latest features and improvements.",{"text":1223,"config":1224},"Read the latest",{"href":1225,"dataGaName":1226,"dataGaLocation":962},"/whats-new/","whats new",{"text":1228,"config":1229,"menu":1231},"Company",{"dataNavLevelOne":1230},"company",{"type":1013,"columns":1232},[1233],{"items":1234},[1235,1240,1246,1248,1253,1258,1263,1268,1273,1278],{"text":1236,"config":1237},"About",{"href":1238,"dataGaName":1239,"dataGaLocation":962},"/company/","about",{"text":1241,"config":1242,"footerGa":1245},"Jobs",{"href":1243,"dataGaName":1244,"dataGaLocation":962},"/jobs/","jobs",{"dataGaName":1244},{"text":1207,"config":1247},{"href":1209,"dataGaName":1210,"dataGaLocation":962},{"text":1249,"config":1250},"Leadership",{"href":1251,"dataGaName":1252,"dataGaLocation":962},"/company/team/e-group/","leadership",{"text":1254,"config":1255},"Handbook",{"href":1256,"dataGaName":1257,"dataGaLocation":962},"https://handbook.gitlab.com/","handbook",{"text":1259,"config":1260},"Investor relations",{"href":1261,"dataGaName":1262,"dataGaLocation":962},"https://ir.gitlab.com/overview/default.aspx","investor relations",{"text":1264,"config":1265},"Trust Center",{"href":1266,"dataGaName":1267,"dataGaLocation":962},"/security/","trust center",{"text":1269,"config":1270},"AI Transparency Center",{"href":1271,"dataGaName":1272,"dataGaLocation":962},"/ai-transparency-center/","ai transparency center",{"text":1274,"config":1275},"Newsletter",{"href":1276,"dataGaName":1277,"dataGaLocation":962},"/company/contact/#contact-forms","newsletter",{"text":1279,"config":1280},"Press",{"href":1281,"dataGaName":1282,"dataGaLocation":962},"/press/","press",{"text":1284,"config":1285,"menu":1286},"Contact us",{"dataNavLevelOne":1230},{"type":1013,"columns":1287},[1288],{"items":1289},[1290,1295,1300],{"text":1291,"config":1292},"Talk to sales",{"href":1293,"dataGaName":1294,"dataGaLocation":962},"/sales/","talk to sales",{"text":1296,"config":1297},"Support portal",{"href":1298,"dataGaName":1299,"dataGaLocation":962},"https://support.gitlab.com/hc/en-us","support portal",{"text":1301,"config":1302},"Customer portal",{"href":1303,"dataGaName":1304,"dataGaLocation":962},"https://customers.gitlab.com/customers/sign_in/","customer portal",{"close":1306,"login":1307,"suggestions":1314},"Close",{"text":1308,"link":1309},"To search repositories and projects, login to",{"text":1310,"config":1311},"gitlab.com",{"href":976,"dataGaName":1312,"dataGaLocation":1313},"search login","search",{"text":1315,"default":1316},"Suggestions",[1317,1319,1323,1325,1329,1333],{"text":993,"config":1318},{"href":998,"dataGaName":993,"dataGaLocation":1313},{"text":1320,"config":1321},"Code Suggestions (AI)",{"href":1322,"dataGaName":1320,"dataGaLocation":1313},"/solutions/code-suggestions/",{"text":1029,"config":1324},{"href":1031,"dataGaName":1029,"dataGaLocation":1313},{"text":1326,"config":1327},"GitLab on AWS",{"href":1328,"dataGaName":1326,"dataGaLocation":1313},"/partners/technology-partners/aws/",{"text":1330,"config":1331},"GitLab on Google Cloud",{"href":1332,"dataGaName":1330,"dataGaLocation":1313},"/partners/technology-partners/google-cloud-platform/",{"text":1334,"config":1335},"Why GitLab?",{"href":1006,"dataGaName":1334,"dataGaLocation":1313},{"freeTrial":1337,"mobileIcon":1342,"desktopIcon":1347,"secondaryButton":1350},{"text":1338,"config":1339},"Start free trial",{"href":1340,"dataGaName":967,"dataGaLocation":1341},"https://gitlab.com/-/trials/new/","nav",{"altText":1343,"config":1344},"Gitlab Icon",{"src":1345,"dataGaName":1346,"dataGaLocation":1341},"https://res.cloudinary.com/about-gitlab-com/image/upload/v1758203874/jypbw1jx72aexsoohd7x.svg","gitlab icon",{"altText":1343,"config":1348},{"src":1349,"dataGaName":1346,"dataGaLocation":1341},"https://res.cloudinary.com/about-gitlab-com/image/upload/v1758203875/gs4c8p8opsgvflgkswz9.svg",{"text":1351,"config":1352},"Get Started",{"href":1353,"dataGaName":1354,"dataGaLocation":1341},"https://gitlab.com/-/trial_registrations/new?glm_source=about.gitlab.com/get-started/","get started",{"freeTrial":1356,"mobileIcon":1360,"desktopIcon":1362},{"text":1357,"config":1358},"Learn more about GitLab Duo",{"href":998,"dataGaName":1359,"dataGaLocation":1341},"gitlab duo",{"altText":1343,"config":1361},{"src":1345,"dataGaName":1346,"dataGaLocation":1341},{"altText":1343,"config":1363},{"src":1349,"dataGaName":1346,"dataGaLocation":1341},{"button":1365,"mobileIcon":1370,"desktopIcon":1372},{"text":1366,"config":1367},"/switch",{"href":1368,"dataGaName":1369,"dataGaLocation":1341},"#contact","switch",{"altText":1343,"config":1371},{"src":1345,"dataGaName":1346,"dataGaLocation":1341},{"altText":1343,"config":1373},{"src":1374,"dataGaName":1346,"dataGaLocation":1341},"https://res.cloudinary.com/about-gitlab-com/image/upload/v1773335277/ohhpiuoxoldryzrnhfrh.png",{"freeTrial":1376,"mobileIcon":1381,"desktopIcon":1383},{"text":1377,"config":1378},"Back to pricing",{"href":1111,"dataGaName":1379,"dataGaLocation":1341,"icon":1380},"back to pricing","GoBack",{"altText":1343,"config":1382},{"src":1345,"dataGaName":1346,"dataGaLocation":1341},{"altText":1343,"config":1384},{"src":1349,"dataGaName":1346,"dataGaLocation":1341},{"title":1386,"titleMobile":1387,"button":1388,"config":1393},"Duo Agent Platform delivers 400% ROI, per new Forrester Consulting study.","400% ROI: Forrester TEI for GitLab Duo",{"text":1004,"config":1389},{"href":1390,"dataGaName":1391,"dataGaLocation":1392},"https://about.gitlab.com/blog/gitlab-duo-agent-platform-delivers-400-percent-roi/","forrester-tei-dap-banner","global-banner",{"layout":1394,"disabled":941},"release",{"data":1396},{"text":1397,"source":1398,"edit":1404,"contribute":1409,"config":1414,"items":1419,"minimal":1627},"Git is a trademark of Software Freedom Conservancy and our use of 'GitLab' is under license",{"text":1399,"config":1400},"View page source",{"href":1401,"dataGaName":1402,"dataGaLocation":1403},"https://gitlab.com/gitlab-com/marketing/digital-experience/about-gitlab-com/","page source","footer",{"text":1405,"config":1406},"Edit this page",{"href":1407,"dataGaName":1408,"dataGaLocation":1403},"https://gitlab.com/gitlab-com/marketing/digital-experience/about-gitlab-com/-/blob/main/content/","web ide",{"text":1410,"config":1411},"Please contribute",{"href":1412,"dataGaName":1413,"dataGaLocation":1403},"https://gitlab.com/gitlab-com/marketing/digital-experience/about-gitlab-com/-/blob/main/CONTRIBUTING.md/","please contribute",{"twitter":1415,"facebook":1416,"youtube":1417,"linkedin":1418},"https://twitter.com/gitlab","https://www.facebook.com/gitlab","https://www.youtube.com/channel/UCnMGQ8QHMAnVIsI3xJrihhg","https://www.linkedin.com/company/gitlab-com",[1420,1467,1519,1563,1595],{"title":1109,"links":1421,"subMenu":1436},[1422,1426,1431],{"text":1423,"config":1424},"View plans",{"href":1111,"dataGaName":1425,"dataGaLocation":1403},"view plans",{"text":1427,"config":1428},"Why Premium?",{"href":1429,"dataGaName":1430,"dataGaLocation":1403},"/pricing/premium/","why premium",{"text":1432,"config":1433},"Why Ultimate?",{"href":1434,"dataGaName":1435,"dataGaLocation":1403},"/pricing/ultimate/","why ultimate",[1437],{"title":1438,"links":1439},"Contact Us",[1440,1443,1445,1447,1452,1457,1462],{"text":1441,"config":1442},"Contact sales",{"href":1293,"dataGaName":972,"dataGaLocation":1403},{"text":1296,"config":1444},{"href":1298,"dataGaName":1299,"dataGaLocation":1403},{"text":1301,"config":1446},{"href":1303,"dataGaName":1304,"dataGaLocation":1403},{"text":1448,"config":1449},"Status",{"href":1450,"dataGaName":1451,"dataGaLocation":1403},"https://status.gitlab.com/","status",{"text":1453,"config":1454},"Terms of use",{"href":1455,"dataGaName":1456,"dataGaLocation":1403},"/terms/","terms of use",{"text":1458,"config":1459},"Privacy statement",{"href":1460,"dataGaName":1461,"dataGaLocation":1403},"/privacy/","privacy statement",{"text":1463,"config":1464},"Cookie preferences",{"dataGaName":1465,"dataGaLocation":1403,"id":1466,"isOneTrustButton":165},"cookie preferences","ot-sdk-btn",{"title":1009,"links":1468,"subMenu":1476},[1469,1472],{"text":953,"config":1470},{"href":991,"dataGaName":1471,"dataGaLocation":1403},"devsecops platform",{"text":1473,"config":1474},"AI-Assisted Development",{"href":998,"dataGaName":1475,"dataGaLocation":1403},"ai-assisted development",[1477],{"title":1478,"links":1479},"Topics",[1480,1485,1490,1495,1500,1504,1509,1514],{"text":1481,"config":1482},"CICD",{"href":1483,"dataGaName":1484,"dataGaLocation":1403},"/topics/ci-cd/","cicd",{"text":1486,"config":1487},"GitOps",{"href":1488,"dataGaName":1489,"dataGaLocation":1403},"/topics/gitops/","gitops",{"text":1491,"config":1492},"DevOps",{"href":1493,"dataGaName":1494,"dataGaLocation":1403},"/topics/devops/","devops",{"text":1496,"config":1497},"Version Control",{"href":1498,"dataGaName":1499,"dataGaLocation":1403},"/topics/version-control/","version control",{"text":952,"config":1501},{"href":1502,"dataGaName":1503,"dataGaLocation":1403},"/topics/devsecops/","devsecops",{"text":1505,"config":1506},"Cloud Native",{"href":1507,"dataGaName":1508,"dataGaLocation":1403},"/topics/cloud-native/","cloud native",{"text":1510,"config":1511},"AI for Coding",{"href":1512,"dataGaName":1513,"dataGaLocation":1403},"/topics/devops/ai-for-coding/","ai for coding",{"text":1515,"config":1516},"Agentic AI",{"href":1517,"dataGaName":1518,"dataGaLocation":1403},"/topics/agentic-ai/","agentic ai",{"title":1520,"links":1521},"Solutions",[1522,1524,1526,1531,1535,1538,1542,1545,1547,1550,1553,1558],{"text":1053,"config":1523},{"href":1048,"dataGaName":1053,"dataGaLocation":1403},{"text":1042,"config":1525},{"href":1025,"dataGaName":1026,"dataGaLocation":1403},{"text":1527,"config":1528},"Agile development",{"href":1529,"dataGaName":1530,"dataGaLocation":1403},"/solutions/agile-delivery/","agile delivery",{"text":1532,"config":1533},"SCM",{"href":1038,"dataGaName":1534,"dataGaLocation":1403},"source code management",{"text":1481,"config":1536},{"href":1031,"dataGaName":1537,"dataGaLocation":1403},"continuous integration & delivery",{"text":1539,"config":1540},"Value stream management",{"href":1081,"dataGaName":1541,"dataGaLocation":1403},"value stream management",{"text":1486,"config":1543},{"href":1544,"dataGaName":1489,"dataGaLocation":1403},"/solutions/gitops/",{"text":1091,"config":1546},{"href":1094,"dataGaName":1095,"dataGaLocation":1403},{"text":1548,"config":1549},"Small business",{"href":1100,"dataGaName":1101,"dataGaLocation":1403},{"text":1551,"config":1552},"Public sector",{"href":1106,"dataGaName":1107,"dataGaLocation":1403},{"text":1554,"config":1555},"Education",{"href":1556,"dataGaName":1557,"dataGaLocation":1403},"/solutions/education/","education",{"text":1559,"config":1560},"Financial services",{"href":1561,"dataGaName":1562,"dataGaLocation":1403},"/solutions/finance/","financial services",{"title":1114,"links":1564},[1565,1567,1569,1571,1574,1576,1579,1581,1583,1585,1587,1589,1591,1593],{"text":1127,"config":1566},{"href":1129,"dataGaName":1130,"dataGaLocation":1403},{"text":1132,"config":1568},{"href":1134,"dataGaName":1135,"dataGaLocation":1403},{"text":1137,"config":1570},{"href":1139,"dataGaName":1140,"dataGaLocation":1403},{"text":1142,"config":1572},{"href":1144,"dataGaName":1573,"dataGaLocation":1403},"docs",{"text":1165,"config":1575},{"href":1167,"dataGaName":1168,"dataGaLocation":1403},{"text":1577,"config":1578},"What's new",{"href":1225,"dataGaName":1226,"dataGaLocation":1403},{"text":1160,"config":1580},{"href":1162,"dataGaName":1163,"dataGaLocation":1403},{"text":1179,"config":1582},{"href":1181,"dataGaName":1182,"dataGaLocation":1403},{"text":1187,"config":1584},{"href":1189,"dataGaName":1190,"dataGaLocation":1403},{"text":1192,"config":1586},{"href":1194,"dataGaName":1195,"dataGaLocation":1403},{"text":1197,"config":1588},{"href":1199,"dataGaName":1200,"dataGaLocation":1403},{"text":1202,"config":1590},{"href":1204,"dataGaName":1205,"dataGaLocation":1403},{"text":1207,"config":1592},{"href":1209,"dataGaName":1210,"dataGaLocation":1403},{"text":1212,"config":1594},{"href":1214,"dataGaName":1215,"dataGaLocation":1403},{"title":1228,"links":1596},[1597,1599,1601,1603,1605,1607,1611,1616,1618,1620,1622],{"text":1236,"config":1598},{"href":1238,"dataGaName":1230,"dataGaLocation":1403},{"text":1241,"config":1600},{"href":1243,"dataGaName":1244,"dataGaLocation":1403},{"text":1249,"config":1602},{"href":1251,"dataGaName":1252,"dataGaLocation":1403},{"text":1254,"config":1604},{"href":1256,"dataGaName":1257,"dataGaLocation":1403},{"text":1259,"config":1606},{"href":1261,"dataGaName":1262,"dataGaLocation":1403},{"text":1608,"config":1609},"Sustainability",{"href":1610,"dataGaName":1608,"dataGaLocation":1403},"/sustainability/",{"text":1612,"config":1613},"Diversity, inclusion and belonging (DIB)",{"href":1614,"dataGaName":1615,"dataGaLocation":1403},"/diversity-inclusion-belonging/","Diversity, inclusion and belonging",{"text":1264,"config":1617},{"href":1266,"dataGaName":1267,"dataGaLocation":1403},{"text":1274,"config":1619},{"href":1276,"dataGaName":1277,"dataGaLocation":1403},{"text":1279,"config":1621},{"href":1281,"dataGaName":1282,"dataGaLocation":1403},{"text":1623,"config":1624},"Modern Slavery Transparency Statement",{"href":1625,"dataGaName":1626,"dataGaLocation":1403},"https://handbook.gitlab.com/handbook/legal/modern-slavery-act-transparency-statement/","modern slavery transparency statement",{"items":1628},[1629,1632,1635],{"text":1630,"config":1631},"Terms",{"href":1455,"dataGaName":1456,"dataGaLocation":1403},{"text":1633,"config":1634},"Cookies",{"dataGaName":1465,"dataGaLocation":1403,"id":1466,"isOneTrustButton":165},{"text":1636,"config":1637},"Privacy",{"href":1460,"dataGaName":1461,"dataGaLocation":1403},[1639],{"id":1640,"title":7,"body":940,"config":1641,"content":1643,"description":940,"extension":1647,"meta":1648,"navigation":165,"path":1649,"seo":1650,"stem":1651,"__hash__":1652},"blogAuthors/en-us/blog/authors/william-arias.yml",{"template":1642},"BlogAuthor",{"name":7,"config":1644},{"headshot":1645,"ctfId":1646},"https://res.cloudinary.com/about-gitlab-com/image/upload/v1749667549/Blog/Author%20Headshots/warias-headshot.jpg","warias","yml",{},"/en-us/blog/authors/william-arias",{},"en-us/blog/authors/william-arias","1h59SugLZ7hePm0SChSE5WG0Z3uurYxGrujcXoxs4tA",[1654,1662,1670],{"title":1655,"description":1656,"heroImage":1657,"category":936,"date":1658,"authors":1659,"slug":1661,"externalUrl":940},"How I built a demo generator with GitLab Duo Agent Platform","Demos haven't always looked like this. Looking back over my career, here's how they have evolved. Follow the tutorial to build your own demo and agent.","https://res.cloudinary.com/about-gitlab-com/image/upload/v1765809212/noh0mdfn9o94ry9ykura.png","2026-08-13",[1660],"Itzik Gan Baruch","agentic-click-through-demo",{"title":1663,"description":1664,"heroImage":1665,"category":936,"date":1666,"authors":1667,"slug":1669,"externalUrl":940},"How to govern agentic AI, MCPs, and AI code assistants","AI agents can write, push, and deploy without human review at each step. Learn a practical framework for controlling access, permissions, and audit trails.","https://res.cloudinary.com/about-gitlab-com/image/upload/v1785337625/ifzlc2hyzaxhzmjgyymh.png","2026-07-31",[1668],"Julie Griffin","govern-agentic-ai-mcps-code-assistants",{"title":1671,"description":1672,"heroImage":1673,"category":936,"date":1674,"authors":1675,"slug":1677,"externalUrl":940},"Claude Opus 5 on GitLab: Reasoning built for the hard tasks","Anthropic’s Claude Opus 5 is now available on GitLab Duo Agent Platform, delivering deeper reasoning without sacrificing speed.","https://res.cloudinary.com/about-gitlab-com/image/upload/v1750099203/Blog/Hero%20Images/Blog/Hero%20Images/blog-image-template-1800x945%20%2820%29_2bJGC5ZP3WheoqzlLT05C5_1750099203484.png","2026-07-27",[1676],"Brittany Lutz","claude-opus-5-on-gitlab-duo-agent-platform",{"promotions":1679},[1680,1693,1705,1717],{"id":1681,"categories":1682,"header":1683,"text":1684,"button":1685,"image":1690},"ai-modernization",[936],"Is AI achieving its promise at scale?","Quiz will take 5 minutes or less",{"text":1686,"config":1687},"Get your AI maturity score",{"href":1688,"dataGaName":1689,"dataGaLocation":1168},"/assessments/ai-modernization-assessment/","modernization assessment",{"config":1691},{"src":1692},"https://res.cloudinary.com/about-gitlab-com/image/upload/v1772138786/qix0m7kwnd8x2fh1zq49.png",{"id":1694,"categories":1695,"header":1697,"text":1684,"button":1698,"image":1702},"devops-modernization",[1696,1503],"product","Are you just managing tools or shipping innovation?",{"text":1699,"config":1700},"Get your DevOps maturity score",{"href":1701,"dataGaName":1689,"dataGaLocation":1168},"/assessments/devops-modernization-assessment/",{"config":1703},{"src":1704},"https://res.cloudinary.com/about-gitlab-com/image/upload/v1772138785/eg818fmakweyuznttgid.png",{"id":1706,"categories":1707,"header":1709,"text":1684,"button":1710,"image":1714},"security-modernization",[1708],"security","Are you trading speed for security?",{"text":1711,"config":1712},"Get your security maturity score",{"href":1713,"dataGaName":1689,"dataGaLocation":1168},"/assessments/security-modernization-assessment/",{"config":1715},{"src":1716},"https://res.cloudinary.com/about-gitlab-com/image/upload/v1772138786/p4pbqd9nnjejg5ds6mdk.png",{"id":1718,"paths":1719,"header":1722,"text":1723,"button":1724,"image":1729},"github-azure-migration",[1720,1721],"migration-from-azure-devops-to-gitlab","integrating-azure-devops-scm-and-gitlab","Is your team ready for GitHub's Azure move?","GitHub is already rebuilding around Azure. Find out what it means for you.",{"text":1725,"config":1726},"See how GitLab compares to GitHub",{"href":1727,"dataGaName":1728,"dataGaLocation":1168},"/compare/gitlab-vs-github/github-azure-migration/","github azure migration",{"config":1730},{"src":1704},{"header":1732,"blurb":1733,"button":1734,"secondaryButton":1739},"Start building faster today","See what your team can do with the intelligent orchestration platform for DevSecOps.\n",{"text":1735,"config":1736},"Get your free trial",{"href":1737,"dataGaName":967,"dataGaLocation":1738},"https://gitlab.com/-/trial_registrations/new?glm_content=default-saas-trial&glm_source=about.gitlab.com/","feature",{"text":1441,"config":1740},{"href":1293,"dataGaName":972,"dataGaLocation":1738},1786803755545]