SmartHQ
Feb 2026 — Apr 2026Turns raw bank statement exports into automated financial insights and a grounded Q&A assistant.
- Python
- FastAPI
- Pandas
- React
- RAG
- Engineered a 7-stage Pandas pipeline transforming transaction data into 16 JSON datasets powering 14 dashboard visualizations.
- Implemented salary and recurring-payment detection using statistical scoring across multi-month transaction patterns.
- Designed a retrieval-augmented Q&A system ranking transactions by merchant, category, and recency before grounding LLM responses — cutting context payload 94% (1,500 → 80 rows).