DevSecOps engineer in Palo Alto, California. I build and run Kubernetes platforms, CI/CD pipelines and machine learning infrastructure.

What I work on:

• Kubernetes platform: Helm overlays for applications deployed with Argo CD on RKE2 and GKE, and the autoscaling, ingress and Terraform-managed cloud resources around them.
• CI/CD: GitLab CI pipelines that build and scan container images and produce an SBOM for each one, and a self-service pipeline for database schema changes.
• ML infrastructure: a FastAPI service that runs multi-stage ML pipelines in containers, and automation for serving large language models with vLLM on GPU nodes.

My day-to-day work is in private repositories. The projects here are personal. The first two are tested end to end by GitHub Actions on a Kubernetes cluster that each run creates:

• inference-on-kubernetes: an inference-style API with a Helm chart, Prometheus error-budget alerts, a Grafana dashboard and KEDA autoscaling. GitHub Actions creates a kind cluster on two Kubernetes versions and tests rollouts, node drains, scaling and alerting under load. The model is a stand-in.
• signed-image-pipeline: a GitHub Actions pipeline that scans a container image, writes an SBOM, then signs and attests it with keyless cosign, plus a Kyverno policy that admits nothing else. Every release tests nine accept and reject cases on a kind cluster.
• rocket-telemetry-pipeline: a data pipeline over synthetic rocket engine telemetry with a Python generator and cleaner, dbt models for Redshift and a Streamlit dashboard, checked by pytest and dbt parse in GitHub Actions.
• nba-game-prediction: a Prefect flow that lands the live NBA scoreboard in S3, a loader into Snowflake, dbt models over the raw JSON, scikit-learn notebooks that backtest win prediction, and a React app that shows bookmaker favourites.
• finance-tracker: a Streamlit dashboard for analysing a stock, comparing tickers and reviewing a portfolio.

Before this I was a data engineer working with Snowflake, dbt and GitLab CI. M.S. Computer Science, Cal Poly Pomona. B.S. Computational Mathematics, UCLA.

LinkedIn · shubhroses.github.io