TAMS — Travel & Meeting Society · Remote
Backend Engineer
Built an auditable multi-agent hotel procurement pipeline backed by Python, FastAPI, PostgreSQL, and structured analytics. We presented the system at the TAMS conference at Harvard University to an audience that included teams from Kayak, Navan, IrisAI, and other major travel companies.
- Built a multi-agent pipeline spanning Discovery, Evaluation, Pricing, and Explanation, with auditable decision traces backed by a Python/FastAPI REST API.
- Converted raw spend data into procurement signals with Pandas cohort analysis, IQR confidence bands, and Amadeus GDS market data.
- Led a five-dimension vendor scoring framework covering price, location, program fit, compliance, and service quality using Pydantic v2 schemas and weighted scoring.
- Built document ingestion and PostgreSQL pipelines that normalized varied RFP responses and efficiently retrieved thousands of historical quotes.
2 weeksRFP lifecycle, down from 6–12 months
10k–50kspend-data rows analyzed per cycle
3–10shortlisted vendors per cycle