MLOps Engineer

Responsibilities

• Design multi-agent architectures, defining agent roles, toolboxes, handoffs, memory strategies, and supervisor policies for safe collaboration.
• Build high-quality Retrieval Augmented Generation (RAG) systems, including ingestion, chunking, embeddings, indexing, retrieval, evaluation (precision/recall, groundedness, hallucination checks), guardrails, and citations.
• Productionize solutions on AWS using services such as Bedrock (Agents/Knowledge Bases/Flows), Lambda, API Gateway, S3, DynamoDB, OpenSearch/Vector DB, Step Functions, and CloudWatch for tracing and alerts.
• Implement MLOps/LLMOps best practices, including automating CI/CD (GitOps), containerization (Docker/Kubernetes), infrastructure-as-code, secrets/IAM management, blue-green deployments/rollbacks, and data/feature pipelines.
• Establish observability and evaluation mechanisms, instrumenting telemetry (traces, token/cost, latency), building dashboards (Grafana/CloudWatch), and integrating human-in-the-loop review, A/B testing, and continuous offline/online evaluations.
• Ensure reliable operation at scale by implementing caching, rate-limit management, queueing, idempotency, and backoff strategies, and proactively detecting drift and degradation.
• Collaborate and communicate effectively with infrastructure, DevOps, data, and architecture teams. Document designs, SLIs/SLOs, and runbooks, and present status and insights to technical and non-technical stakeholders.