I build full-stack AI products that combine production web development with reliable LLM workflows.

Instead of simply connecting an application to an AI API, I build the surrounding system needed to make AI features useful in a real product.

I can help you build:

• AI-powered SaaS applications • Agentic workflows with LangGraph • RAG and semantic search systems • AI agents and tool-calling workflows • OpenAI and Gemini integrations • Multimodal AI workflows using text and images • Embedding and vector-search pipelines • Supabase and pgvector integrations • Structured LLM outputs and validation • Retry and conditional workflow logic • Next.js frontends and APIs • PostgreSQL-backed applications • Authentication and user-specific AI experiences

My recent work includes VibeFit, an agentic AI wardrobe platform built with Next.js, LangGraph, OpenAI, Gemini, Supabase, and pgvector.

I designed independent ingestion and retrieval workflows that transform clothing images into structured metadata and embeddings, store them for semantic search, and retrieve relevant garments before generating recommendations.

The system uses validation, structured outputs, retry strategies, conditional routing, and observability to make the AI pipeline more reliable and reduce hallucinations.

If you have an AI product idea, existing prototype, or workflow you want to turn into a usable application, I can help build the system from frontend to AI orchestration and data layer.