Hands-On Engineering Projects

Build an end-to-end RAG system with document ingestion, chunking, hybrid vector retrieval, and reranking.

Implement a compliant MCP server exposing custom databases, filesystem tools, and schema validation.

Construct a pure Python ReAct agent execution engine with memory management, tool execution, and retry mechanisms.

Build a stateful multi-agent DAG for collaborative research, drafting, reviewing, and code verification.

Construct an automated test harness measuring hallucination rates, RAG faithfulness, and latency budgets.

Package an inference model into a hardened, multi-stage CPU/GPU Docker container with Prometheus metrics.

Full-scale deployment combining RAG, tool calling, multi-agents, and observability into one verified platform.

Build an autonomous self-healing agent platform with Redis Streams buffering, LangGraph cyclic state machine, MCP tool contracts, OPA guardrails, and 120s synthetic health check with automated rollback.