I build
Lead GenAI Engineer at SortUp. I take AI products from zero to production — agents, RAG pipelines, MCP tooling, document intelligence. Whether you need someone to build your system, collaborate on architecture, or help you think through a hard problem — let's talk.
Background
Most AI projects are demos. I build the ones that keep running after the client calls end — with proper orchestration, evals, and the plumbing to hold it all together in the real world.
My core work lives at the intersection of multi-agent systems, RAG architecture, and document intelligence. I've fine-tuned models for construction ERP extraction, shipped voice agents for medical transcription, and built MCP-based automation for live booking systems.
I can take your product from a whiteboard idea to a deployed, production-ready system — fast. Whether you need a technical co-thinker for early-stage scoping or a senior engineer to own delivery end-to-end, I'm available for both.
Selected Work
All production systems — real clients, real traffic, real edge cases.
Multi-model AI for the funeral industry. Guides families through structured story collection and generates polished obituaries with tone adjustment, fact checks, and revision flows.
Personalized AI recommendation system built on pgvector and Qwen. Uses MMR (Maximal Marginal Relevance) search for diversity-aware suggestions that avoid repetitive results.
Fine-tuned GPT-4o-mini for construction PDF extraction. Ingests unstructured estimate documents and outputs ERP-ready structured JSON — with a weekly retraining pipeline that adapts automatically.
Autonomous self-healing AI agent for network troubleshooting. Detects and resolves faults without human intervention — built on LangGraph + ReAct, monitored via Grafana.
AI-powered content intelligence platform for product innovation. Ingests documents, extracts insights, and generates structured content across multiple formats using multi-model routing.
Authenticated Islamic advisory chatbot with RAG over curated religious texts. Provides reliable, source-grounded answers to spiritual and legal queries with full citation traceability.
RAG-powered chatbot platform. Users upload PDFs or point to any URL and instantly get a context-aware assistant they can embed on their own website.
Conversational clinical agent that analyzes lab reports, performs differential diagnosis, and recommends follow-up tests. Voice-enabled via VAPI, powered by MCP + PydanticAI.
AI pipeline that transcribes medical consultations and outputs dual structured reports — a clinical note for doctors and a plain-language summary for patients — with specialty-specific prompt tracks.
AI-native analytics platform. Centralizes business metrics into a single dashboard with natural language querying — non-technical teams interrogate their own data without writing SQL.
Fully automated appointment system for pet stores — connects GHL and MoeGo via MCP. Handles scheduling, confirmations, and follow-ups end-to-end with zero human touch.
AI resume screener that ingests a job description and batch of CVs, then scores and ranks candidates automatically — cutting initial screening from hours to seconds.
Ready to develop your product, collaborate on architecture, or just help you think through a hard problem. If you have a GenAI challenge worth solving — I'm one call away.