Data Science analyst & Mentor
M.Sc CS in A.I and M.L
DS
Projects
PortFolio of Computer Vision and Nlp Projects
and
Masters - Business Case study using ML
3d Object Detection and Tracking of Lidar Point Cloud using POINT-RCNN ,Casa and 3d-MOT Tracker
• Developed Point-RCNN for precise 3d object detection and object tracking in LIDAR Point Clouds.
• Conducted rigorous benchmarking of 3 different Object Detectors (Point-RCNN, Cas-A, and PV-RCNN)., achieving over 20 FPS processing speed.
• Attained a remarkable 95% recognition rate for vehicles, including accurate distance and velocity estimation.
• Engineered a custom bounding box visualizer, for enabling comprehensive analysis and visualization of detected objects and tracking in 3d
LLM BASED DOC-CHAT ASSISTANT
• Orchestrated hosting and administration of an Active Loop Vector Database, with 40% Efficient storage and retrieval of high-dimensional Multi-format documents with 100% secure storage of sensitive documents.
• Engineered a user-centric Streamlit web application, enabling seamless document retrieval from the hosted database.
• Integrated LangChain and OpenAI GPT models within the application to Provide Retrieval-Augmented Generation (RAG) Finetune prompting for the Document
• Employed MultipleNegativesRankingLoss methodology to rigorously validate LLM output queries from retrieved documents against original answers and Successfully achieved a stable loss metric of 0.55