Python · Django · FastAPI · RAG · Agents · MCP · AWS · Bedrock
I ship production AI into real products — multi-tenant logistics, healthcare analytics, and LLM systems (RAG, SQL-over-data, voice agents).
| Focus | Details |
|---|---|
| LLM apps | RAG, Bedrock/Claude, SQL-over-data chatbots, MCP tool servers |
| Voice & agents | Retell AI agents, exception call/email flows, workflow AI nodes |
| Backend | Django/DRF, FastAPI, Celery, Kafka microservices on AWS |
| Data & ML | Postgres RLS, ETA models, alert pipelines, multi-tenant analytics |
Real-time life-sciences platform: shipments, IoT sensors, inventory, GxP compliance, maps, alerts, and AI exception handling.
- Insights chatbot — Claude on Bedrock → validated SELECT-only SQL under Postgres RLS (multi-tenant)
- Voice + MCP — Retell agents with live tools (shipments, alerts, sensors, CoC, routes)
- AI interventions — call/email exception handling, Logiblocks AI nodes, transcripts on alerts/CoC
- Product — Temperature GxP, inventory, Control Tower, home KPIs, live maps, alert system
- Stack — Django 5 · React/TypeScript · Kafka · ECS/Lambda · AWS
Private company code — skills mirrored in public repos below.
Node-based Python platform (Flask ETL → FastAPI) with backend-owned forms and schemas.
- Cohort filters & operator validation for member segmentation
- Quote disposition APIs · coversheet eligibility patching
- Employer/member savings engines → recommendation nodes
- Claims analytics fixes (paid amounts, age typing)
| Repo | Focus |
|---|---|
ai-interviewer |
LLM interview agent |
document-qa-rag |
RAG / document Q&A |
django-backend |
Production Python API patterns |
uv · pre-commit (Black / Ruff / isort) · mypy · pytest / Vitest · Taskfile gates before PR
Typed Python, safe SQL, clear service boundaries — not notebook prototypes.
