romilan24 / diegocalderon.github.io

Diego Calderon's data science portfolio.

Diego Calderon

Experienced engineer specializing in forecasting, machine learning, deep learning, and optimization in Energy and Finance. My current interests include Multi-Agent LLM networks, AI detection algorithms, and Vector databases.

Follow for more and let me know if you would like to collaborate on any projects.

📋 Project To Do list:

Energy

• Solar, Wind, Storage capacity forecasts (will use Open Meteo API https://github.com/romilan24/multi-param-open-meteo)
• Pull AS prices from the CAISO OASIS API
• Energy + Ancillary Service co-optimization (will base on https://github.com/romilan24/Battery-Storage-Optimization-Strategy)
• Time Series forecast w/ Mamba models
• High frequency state space models (Kalman filter) for Real-time price forecasts

AI / ML / DL

• Streamlit webapp for Ollama LLM inference
• Turing Test => AI vs Human detection algorithm (BERT or GPTZero)
• Reinforecement Learning applied to LLM performance
• Whipser + LLM app (Rhasspy Piper + Llama 3)
• Multi-Agent LLM systems:
  • Optimal prompt design assistant; describe what you want to agent who builds optimal prompt for you
  • Coding Agent + python kernel + Streamlit UI; tell Coding assistant what code you want and they build it, test it for you, fix any mistakes and give you final code
• LoRA app and/or RAG application (Langchain or Llama Index, Qdrant or Chroma vdb)
• Document type sorting (https://github.com/google/magika)
• Automated Self-Driving, traffic prediction

💻 Tech Stack:

🌐 Let's Connect:

📧 [email protected] • 📍 Los Angeles, CA USA