I build production-grade applications across mobile, backend APIs, data pipelines, and AI systems. I design the architecture, ship the product and own the quality end to end.
A full-stack React Native app that scans food barcodes and delivers personalised ingredient analysis against a user's health profile.
- Stack: React Native (Expo) · TypeScript · FastAPI · Supabase · Groq LLM · Railway
- Live backend: https://web-production-bab72.up.railway.app/health
- Architecture: RAG pipeline · JWT auth · RLS · Sentry monitoring · GitHub Actions CI/CD
- Privacy-first: All health data stored locally on device — zero server-side health data
An MCP server connecting Claude Desktop to enterprise vendor risk data via 6 AI-powered tools.
- Stack: FastMCP · Pinecone · Gemini Embeddings · LangGraph · Groq · Supabase
- Capabilities: RAG compliance analysis · Semantic vector search · Stateful risk delta agent · News intelligence
A production data pipeline tracking social sentiment across 6 alcohol brands in real time.
- Stack: Python · DLT · dbt · Railway Postgres · FastAPI · Pinecone · Gemini Embeddings
- Live: https://spiritpulse-production.up.railway.app/dashboard
- Pipeline: DLT ingestion → dbt transformation → FastAPI → Pinecone vector search
4 Claude agents scoring supplier sustainability reports against GRI, TCFD, and CSRD in under 90 seconds.
- Stack: Claude · Multi-agent orchestration · ESG framework analysis
Mobile & Frontend: React Native · Expo · TypeScript
Backend: FastAPI (Python) · REST APIs · Docker
AI & LLM: Groq · RAG Pipelines · Prompt Engineering · Anthropic Claude
Database & Auth: Supabase (PostgreSQL) · Row Level Security · Edge Functions
DevOps: Railway · GitHub Actions · Sentry · Docker
Automation: Make.com · Zapier · n8n · OpenAI · Airtable · Notion
Data Engineering: DLT · dbt · PostgreSQL · SQLAlchemy · FastAPI
Vector Search & RAG: Pinecone · Gemini Embeddings · LangChain · Chunking strategies
AI Agents & MCP: Claude · Groq · LangGraph · MCP servers · Prompt Engineering
- Full-stack mobile apps — end to end from camera to backend to database
- AI-powered APIs — RAG pipelines, grounded LLM summaries, vector search
- Enterprise automation — multi-agent pipelines for legal, compliance, and operations
- Privacy-first architecture — zero breach surface design for health and sensitive data
- Data pipelines — end to end ingestion, transformation with dbt, and vector search with Pinecone
-
- MCP servers — custom tools connecting Claude Desktop to enterprise data systems