Projects
A few projects that capture how I think about LLM systems, evaluation, and applied ML.
Fortune 50 enterprise
Migrated batch ingestion to near-real-time Spark/Kafka pipelines, reducing latency to under 5 minutes while improving throughput by 40% and lowering cost by 15%.
Fortune 50 enterprise
End-to-end RAG forecasting system grounding LLM responses in fresh supply-chain telemetry. Lifted predictive accuracy by 30%.
Fortune 50 enterprise
Offline + online evaluation harness with golden sets, faithfulness and contradiction scoring, and release gating on quality deltas.
Ph.D. Dissertation University of the Cumberlands
Comparative study of summarization and retrieval-augmented generation as memory mechanisms for long-running conversational agents.
M.Sc. Research, University of Kentucky
FTIR spectroscopy coupled with machine learning to detect and quantify gluten contamination in grain-based foods.
Open Source
Reproducible evaluation harness comparing summarization-based memory and retrieval-augmented generation for long-term conversational performance in LLMs (LongBench, LoCoMo, LongMemEval).
Open Source
AI-assisted, local-first engine that automatically maps messy data to your canonical schema — with a confidence score and a plain-English reason for every match. Profiles, validates, and detects schema drift.
Open Source
Open-source CLI and MCP server for scanning business datasets, detecting data quality issues, and generating AI-powered data health reports.
Open Source
Open-source LLM memory governance and evaluation toolkit for tracking, scoring, auditing, and improving conversational memory and context retrieval.
Open Source
Turns written short-fiction episodes into ready-to-post vertical videos for TikTok, Reels & YouTube Shorts — all from your laptop.
