Lead Architect & Developer | Artemis Edge AI

• Engineered and deployed an edge-based predictive modeling system utilizing a custom logistic regression classifier optimized via gradient descent to calculate real-time win probabilities for NBA and NCAA matchups.

• Developed a robust features pipeline incorporating dynamic variables including Elo differentials, rest-day metrics, rolling team form, offensive/defensive ratings, and quantitative injury impact data.

• Implemented a rigorous validation framework featuring automated train/test splits across 5 simulated seasons of historical sports data; measured model efficacy using out-of-sample metrics including Accuracy, AUC-ROC, and Log-Loss to eliminate data leakage and memorization.

• Designed a systematic betting logic engine that isolates market inefficiencies by identifying statistically significant "edges"—discrepancies between the model's squashed sigmoid probabilities and sportsbook implied odds.

• Built an integrated backtesting simulator to replay trading strategies on held-out datasets, successfully projecting historical win rates, Return on Investment (ROI), and compounding equity curves.

• Formulated architecture for scalable future iterations, outlining blueprints for online/incremental learning loops integrating live data feeds, cloud-persisted weight states, and automated cron-scheduled retraining jobs.