Portfolio¶
Now - Senior Platform Engineer, V7¶
Building the platform behind V7 Go, V7's AI agent platform for document-heavy workflows in private equity and finance. Recent and ongoing work:
- Compliance - Terraform-managed AWS/GCP/Azure controls and Secureframe evidence for SOC2, ISO 27001, HIPAA and GovRAMP.
- Databases - Own RDS/Postgres: major-version upgrades, maintenance windows, and on-call cover.
- Identity & access - Auth0 tenants, SSO, SCIM; IAM access controls across AWS, GCP and Azure for all engineers.
- CI/CD - Own the runs-on self-hosted GitHub Actions stack on AWS; all CI runs through it. Drove build-time reductions and automated runner upgrades; currently building a merge-queue / release-train flow for Platform repos. Also own shared reusable workflows, Dependabot rollouts, and Actions supply-chain security (action pinning, runner hardening).
- Preview environments - Owner of Northflank for per-PR preview environments across frontend and backend repos.
- Observability - Own the Grafana stack across all envs: AMG (Amazon Managed Grafana), AMP (Amazon Managed Prometheus), Tempo for traces, alerts-as-code via Terraform, and incident.io for on-call (migrated off PagerDuty) plus the 24/7 V7 Go status page. Site24x7 for synthetic monitoring checks.
- Secrets management - Built the central secrets pipeline using 1Password, Terraform and AWS Secrets Manager.
- Reliability & DR - RTO/RPO validation across V7 Go and V7 Darwin; circuit-breaker protection in production.
- Multi-cloud & regions - stood up V7 Go on GCP (GKE + Argo, GCP Marketplace listing); rolled out V7 Go circuit-breaker protection and Metabase analytics into US and UAE regions.
- Onboarding, interviewing & office hours - mentoring new platform engineers, interviewing platform candidates, weekly platform office hours.
Earlier, I maintained darwin-py,
the open-source Python SDK for V7 Darwin (V7's computer-vision data
platform).
Previously¶
ML & MLOps Engineer, X-Chem (Glamorous-AI) - London Core services, CI/CD across projects, JupyterHub for the team.
SDE - Open Source, Segmind - Bengaluru MLOps platform and open-source Python library. Forked MLflow into "mlflow-lite" for in-notebook experiment tracking.
Full Stack Lead, Gram Power - Jaipur ML for appliance disaggregation (clustering, unsupervised). Government, payments and DevOps integrations.
Full Stack Developer, Thoughts2Binary - Jaipur Django + AWS serverless (Lambda, SQS, S3, DynamoDB). CMS work and payment gateways.
Publication¶
- Human Action Recognition Using Multi-Stream Fusion and Hybrid Deep Networks
- IEEE, 2024
Stack¶
Day-to-day: Python, Terraform, AWS, GCP, PostgreSQL, GitHub Actions, Docker. Comfortable with: Go, Rust, Argo Workflows, Kubernetes, PyTorch.