I don't just build models — I build systems that make decisions.
ML Engineer who takes models from notebook to production. I build end-to-end AI systems — prediction services, RAG pipelines, LLM evaluation frameworks — with observability, explainability, and deployment baked in from day one. Currently focused on LLM reliability, AI agents, and production-grade ML infrastructure.
End-to-end ML system · 88.5% accuracy · FastAPI + Streamlit + SHAP + MLflow + Docker
Production healthcare ML system with SHAP explainability, MLflow experiment tracking, REST API, interactive dashboard, and full Docker deployment. Built for trust, not just performance.
QLoRA fine-tuning · +69% ROUGE-L over base · Live Hugging Face demo
Fine-tuned Microsoft Phi-4 Mini 3.8B on SEC 10-K financial Q&A. Demonstrates practical LLM adaptation for domain-specific enterprise use cases.
Automated prompt regression detection · CI/CD for LLMs
Production evaluation pipeline that catches quality regressions before prompts reach users. CI/CD for LLM behavior — because shipping a broken prompt is just as bad as shipping broken code.
Persistent memory · Semantic search · Multi-tenant · Python + TypeScript SDKs
Memory infrastructure for AI agents with semantic search, multi-tenant isolation, 8 framework adapters, and dual-language SDK support.
Multiple search strategies · Semantic chunking · Real-time document processing
Production-ready RAG pipeline with hybrid retrieval, semantic chunking, and real-time ingestion. Goes beyond naive vector search.
Fine-tuned DistilBERT · 19 categories · Drift detection · React dashboard
NLP classifier with FastAPI serving, real-time drift detection via Evidently AI, and a React analytics dashboard. Fully containerized.
I'm actively looking for ML Engineer and Data Scientist roles where I can build production AI systems end-to-end. Especially interested in teams working on LLM reliability, MLOps infrastructure, or AI-powered products.