I am a Python and backend developer specializing in deterministic AI systems, grounded RAG pipelines, and automated data workflows. I focus on building reliable engineering guardrails around LLMs—preventing hallucinations, enforcing strict schemas, and eliminating silent pipeline failures.
What I build and ship:
• High-Velocity Execution: In my recent contract, I architected and deployed three production internal systems directly to a self-hosted VPS in under a month—including a real-time team chat interface, an OSINT record resolution service, and an autonomous GitHub/Reddit RAG technical document generator.
• Deterministic RAG & Extraction: Built document ingestion and extraction pipelines using hybrid search, Cross-Encoder re-ranking, and downstream spaCy NER verification layers to parse 75+ complex document layouts without hallucinated figures or schema drift.
• Agentic State Workflows (Auto-Resume): Designed multi-step AI pipelines using LangGraph state dictionaries, local embeddings (all-MiniLM-L6-v2), and SQLite step-logging for granular observability and rapid root-cause debugging.
• Backend & Data Ingestion: Built asynchronous web extraction and batch pipelines (FastAPI, Playwright, BeautifulSoup) with strict Pydantic validation checks and automated error-recovery loops.
I value fast-paced remote teams, high shipping velocity, and taking full technical ownership from ingestion to deployment.