At Deloitte, I analyzed enterprise telemetry datasets and used Python, Pandas, and NumPy to cleanse and transform data for business intelligence reporting. My automated dashboards helped senior stakeholders track operational KPIs across four business units.
I designed and led the architecture for Buy or Wait Enterprise Financial Decision Agent, a full-stack financial decision agent that processed test transactions and retrieved external financial data through Model Context Protocol. Its PDF extraction pipelines handled more than 50 document types.
I led development of a local AI pipeline that used Qwen2.5-Coder-3B via Ollama to extract hardware parameters from technical ISA documents. Pydantic v2 data contracts helped ensure valid outputs across the test documents.
I also architected a Retrieval-Augmented Generation pipeline with automated PyTorch-backed CI/CD evaluation, and curated structured datasets for machine learning benchmarks and multi-agent LLM workflows.
