Own the complete MLOps backbone for AI/ML and GenAI solutions, from model packaging and CI/CD delivery through production monitoring. Bridge data-science experimentation and enterprise-scale deployment, and set engineering standards for the Enterprise AI MLOps practice.
Experience:
7+ years in ML/AI engineering or DevOps (including 4+ years of production MLOps)
Location:
Bangalore (Whitefield) - Work from office, 5 days a week
Notice Period:
Immediate joiners preferred (0-15 days)
Education
Master's preferred, or Bachelor's in CS, Software Engineering or Data Science
Please note: An online assessment is mandatory as part of the selection process.
Key Responsibilities
- Design end-to-end pipelines - ingestion, training, evaluation, packaging, versioning, deployment and monitoring
- Own Jenkins CI/CD (Declarative/Scripted) for Dev to QA to Production with multi-stage gates
- Deploy model-serving APIs on AKS (FastAPI, vLLM); apply ONNX/TensorRT and blue/green rollouts
- Build observability stacks - drift detection (Evidently AI), Prometheus/Grafana, Azure Monitor
- Enforce DevSecOps - zero-credential pipelines via Key Vault + Managed Identity, SonarQube, Trivy/Snyk
- Set MLOps engineering standards and mentor engineers across the practice
Must-Have Skills
- 7+ years in ML/AI engineering or DevOps with 4+ years operating production MLOps pipelines
- CI/CD and dev tooling: Jenkins (expert), Azure DevOps, ArgoCD (GitOps); UV, Ruff, Pyrefly, pre-commit
- MLOps and data: Databricks end-to-end (Delta Lake, Workflows, MLflow, Unity Catalog), Airflow, DVC, PySpark
- Serving and infra: FastAPI, ONNX/TensorRT, AKS, Helm, Docker, KEDA, Azure APIM, Terraform
- Databases and storage: PostgreSQL, Redis, ADLS Gen2
- A/B testing and canary/shadow deployment experience
- Proven technical leadership, mentoring and stakeholder communication
Nice to Have
- LLM fine-tuning pipelines and large-scale model-serving optimization
- Ray Serve, BentoML for advanced model serving patterns
- Experiment tracking and governance (W&B / Comet ML, Unity Catalog policies)
- Multi-cloud or hybrid deployment strategy and cost optimization
- Semiconductor or manufacturing domain experience