Experience — where the code runs.
AI Engineer (Contract)
Oct 2026 – Present
Healthcare certification project · Germany (Remote)
- Building an AI system that reviews certification applications by checking uploaded documents against what applicants entered in their forms.
- Caught and fixed a bug where valid certificates scanned with phone apps were read as expired, which would have sent good applications back.
- Fixed a rule that rejected whole applications over one missing file, so the file is requested instead.
- Added a second, independent reading of each document that flags disagreements for human review, and built a 33-case test set including tampered files.
Open Source Contributor
Oct 2026 – Present
IPython · git-cola · icalendar · scikit-learn
- Got fixes merged into IPython (image display) and git-cola (error reporting) within two weeks.
- Have further fixes in review at icalendar and scikit-learn, two widely used Python libraries.
Freelance AI Engineer
2024 – Present
Chemical manufacturer · Hyderabad
- Designed the spreadsheet system the operations team ran their day-to-day work on.
- Built an AI assistant that let staff ask questions about their Excel data in plain English and see the exact cell behind each answer; used daily until the company moved to an ERP.
- Open-sourced the engine behind it as ExcelMCP, an MCP server for live Excel access.
Projects — shipped and public.
- Built an AI agent that reads documents, analyses data and writes Word, Excel and PowerPoint reports.
- Kept it working through API outages by switching automatically between AI providers and a local model.
Python · LangGraph · FastAPI · Docker · OCR · RAG · Ollama
- Gave Claude Code a memory that lasts between sessions, published as a Python package on PyPI.
Python · PyPI · Claude Code
ExcelMCP — live Excel access for AI agents
github ↗
- The open-sourced engine behind the manufacturer’s assistant: 11 MCP tools that answer from live workbook data, with the file, sheet and cell behind every number.
Python · FastMCP · Microsoft Graph API · FastEmbed · Pandas
House Price Prediction — machine learning pipeline
2025
- Benchmarked 11 models and tuned the best to price 79% of houses within 10% of their actual value.
scikit-learn · XGBoost · LightGBM · Pandas · NumPy
