╔═══════════════════════════════════════╗ ║ DAKOTA.CHANG >> ml_projects.py ║ ╚═══════════════════════════════════════╝
Building intelligent systems • Solving real-world problems
Engineered ensemble ML pipeline for autonomous HVAC optimization using Offline RL, VAE, GBR, and LSTM.
Sole ML engineer at Barn Owls Technology. Trained YOLOv11 model achieving 90%+ accuracy for real-time pest detection.
Developed LSTM model to predict soccer player positions with <10m mean absolute error.
Led research team developing methods for benchmarking LLM models. Abstract and poster published at AAAS.
Reduced OCR inference training costs by 12% through DSPy benchmarking at Indico Data.
Built Streamlit-based tool for instant prompt quality feedback, improving ML team productivity at Indico Data.
Co-founding Nuggy to revolutionize cannabis retail with AI-powered dynamic pricing. Developing machine learning solutions to help dispensaries optimize margins and increase profitability.