AI Engineer · AI Systems Architect · Detroit, MI
Sole architect on every system I've shipped since 2022: RAG and agent frameworks, real-time decision systems, forecasting. I own the whole thing: the architecture, the result.
Read the real data before any model. Most AI failures are data failures.
Each step asks the model once, with a tight brief and a cost cap.
Deterministic gates test the output. A failed gate gets one repair, then stops.
Nothing reaches people or production without a person pressing approve.
Every run leaves a receipt: what ran, what it cost, what changed, how to undo it.
Built the Dealer Training Hub and a multi-mode AI assistant (LLMs, RAG, agent workflows) that replaced four spreadsheets and a paper sign-up process.
LSTM forecasting plus a production RAG + agent framework for financial decision support. A real-time audio intelligence platform, running in production.
ECU and vehicle-diagnostics data in Python, SQL, Spark, Hadoop and MATLAB; PCA and clustering surfaced recurring defect patterns and cut ECU issue resolution time by about 20%.
Fraud detection and customer segmentation; clustering and predictive models found about 20% more fraudulent locations, explained to the business with SHAP.
Deep learning across ~250K vehicle configurations; 108 predictors cut to 14 key drivers.
R-Keras demand forecasts for 50+ locations, cardiovascular biostatistics, applied ML research on 100M+ row datasets.