Mayur Mahurkar

Computer Vision Engineer @Adagrad AI

Pune, IN

Developed person fall detector using object detection model with oriented bounding boxes, achieving an accuracy of 90+% and successfully deploying it on NVIDIA Triton Server- Training latest YOLO models for Number Plate Recognition models and dockerised deployment on various NVIDIA’s edge devices such as TX2, Jetson Nano, Orion- Deployed an optimised version of Hugging Face’s VIT model for Skin Segmentation and Not Safe for Work image classification on NVIDIA Triton Server and Fast API- Fine-tuned Qwen: Image-to-text LLM model to describe extensive details of the vehicle captured in the frame- Lead, mentor, and manage a high-performing team of 20 data annotators, fostering a collaborative and results-driven work environment- Extensively build various annotation tools using Streamlit and apps with frontend and backend functionalities using AI-embedded coding assistants such as Cursor- Developed LED indicator module for device status checking saving man hours of field engineers- Core contributor in Data pre-preprocessing strategies and worked on internal scripts to better evaluate models, analysis of input/output data, conversion of input data etc.

Government Institute of Science Nagpur

Master of Science (M.Sc.)

Nagpur University

Bachelor of Science (B.Sc.)

Kendriya Vidyalaya

AISSCE