I build production-ready, neuro-symbolic AI pipelines for InsurTech and enterprise automation. I focus on bridging the gap between non-deterministic LLM extraction and deterministic algorithmic routing.

Neuro-Symbolic AI Underwriting Portal A production-grade microservice that automates complex document analysis.

• Architecture: State machine orchestrated via LangGraph; asynchronous processing with FastAPI BackgroundTasks.
• Extraction: Multimodal OCR ingestion (AWS Textract) coupled with semantic entity extraction.
• Privacy: Localized PII sanitization (spaCy/Microsoft Presidio) to ensure data sovereignty before LLM inference.
• Reasoning: Vector RAG (PostgreSQL/pgvector) to inject ground-truth actuarial rules into the LLM context.
• Routing: Deterministic Python engine for final premium calculation, eliminating hallucination risks.

• Languages: Python (Primary), TypeScript, SQL
• Backend: FastAPI, Asynchronous Task Queues, gRPC
• AI/ML: LLM Orchestration (LangChain/LangGraph), RAG, Vector Databases (pgvector), Prompt Engineering
• DevOps: Docker, AWS (S3, Textract), CI/CD fundamentals

I am currently seeking opportunities to contribute to high-scale AI infrastructure and distributed backend systems.