Lead Data Scientist | Applied AI Systems | GenAI & RAG | Operational AI
Math PhD turned Applied AI Engineer building operational AI systems for complex real-world environments — from industrial signal intelligence to production-grade RAG, agentic workflows, evaluation, and observability.
-
operational-ai-copilot
“Ask an AI copilot why your machine is failing.”
Intent-driven agentic RAG over maintenance tickets, manuals, and sensor data: pgvector + nomic embeddings → 6-intent orchestrator → Ollama LLM. With human-in-the-loop feedback, a 20-question eval CI gate, and Prometheus/Grafana observability out of the box. -
industrial-audio-rag
“Ask natural-language questions about factory sounds.”
Full-stack RAG pipeline over DCASE2024 audio: signal features → embeddings → Qdrant → LLM. With CI, FastAPI, and snapshot-based MLOps. -
LorenzBraids
Dynamical systems meets topology. Converts symbolic orbits of the Lorenz system into braid group diagrams. -
RayTracerHaskell
A clean ray tracer written in Haskell — because geometry + FP is fun. -
MindLattice
Explorations in embedding-based memory systems and vector-based reasoning.
-
meshed (i2mint/meshed)
Top contributor (#2 overall) — 90+ commits. Functional graph pipelines for composable Python workflows. -
front (i2mint/front)
UI generation for Python objects. Contributed to rendering logic. -
streamlitfront (i2mint/streamlitfront)
Auto-generates Streamlit frontends from Python functions and objects. Contributed layout adapters and Streamlit integration logic.
