At DeepTorch, I design and deliver custom AI products for clients, including domain-specific chatbots, automated document processing, and predictive analytics pipelines. My ongoing research focuses on deep reinforcement learning agents for cloud data centers and machine learning model capabilities on circuit networks.
At Ntigra, I led development of an in-house LLM system spanning data engineering, training, validation, and deployment. It replaced external APIs, improved performance by 7%, and reduced operational costs.
I also built a production agentic RAG system for medical ICD coding, using ChromaDB to search more than 250,000 vectors, and developed clinical data agents for verified patient-record operations and diagnostic reasoning. The work included context-aware few-shot retrieval that improved F1 score by 13% over static prompting.
My open-source Trading Signal Server combines crypto market analysis, signal generation, backtesting, and an LLM voting pipeline. I’ve also researched sign language translation at DeepTorch and built machine-learning workflows with Pandas, NumPy, and Scikit-learn during my training at SHAI.
