Lead AI Engineer (Contract)
AI-NativeForward-Deployed AI Delivery
Governed Asset Agents
Embedded with the client to build an Enterprise Asset Management system of governed, multi-agent asset agents that predict, analyse, and act on operational data — combining anomaly detection with LLM reasoning, wrapped in guardrails, Responsible AI, and human-in-the-loop review.
Universal RAG Copilot (MCP)
Engineered a RAG Copilot on foundation models (OpenAI, Claude, Hugging Face, LLaMA) that plugs into any system through a secured Model Context Protocol (MCP) connection, using tool/function calling and semantic search over enterprise data — with evaluation keeping responses grounded.
LLM Document Parser
Built an LLM document parser using embeddings, vector search, and RAG to turn unstructured enterprise data into reliable structured output.
IT Support Copilot
Built a policy-driven, governed RAG copilot for multi-tenant organisations: a governed agentic workflow with a three-tier retrieval cascade (hybrid pgvector + BM25, RRF fusion) on Azure AI Foundry and Copilot Studio, grounded by content filtering and groundedness evaluation, with structured auto-escalation to IT.
Tech Stack