Projects
GenAI Simulation Tools
Created LLM-driven dynamic scenarios for medical simulations using LangChain, Python, and semantic search. Built Python-based feedback loops, force prediction models, and vector pipelines for real-time scenario interaction. Integrated Azure AI Studio for seamless model deployment, enhancing the performance of the simulations.
AI Detect
Developed AI Detect, leveraging PyTorch as its deep learning framework, supported by Python, MMEngine, and MMCV from the OpenMMLab ecosystem. Developed training, evaluation, and model integration pipelines entirely in Python to support 2D and 3D keypoint detection, supporting a wide array of SOTA algorithms, techniques, and backbone architectures. Implemented custom data loaders, augmentations, and export tools for seamless model deployment and testing.
Haptic Simulation Tool
Developed a data-driven haptic simulation tool using iMSTK, Chai3D, and Random Forest for force feedback estimation in medical training. Integrated LangChain and LLMs to dynamically generate clinical scenarios for simulation environments. Designed and deployed a RAG pipeline with OpenAI embeddings and Chroma to ground LLM responses in medical data.