Open to opportunities

Yash Sharma

@yoder2k4

Message

Research intern building on-device computer vision and full-stack products with strong ML and web fundamentals.

India

Message

What I'm looking for

I’m looking for a role where I can build efficient ML/computer-vision systems and ship full-stack features—focused on real-world performance, clean engineering, and strong team collaboration.

I’m a Research & Development Intern who builds efficient, real-world computer vision systems—like deploying a transformer-based image captioning model (BLIP) for real-time performance with ≤2s latency and a ≤200MB memory footprint on low-end mobile CPUs.

I also enjoy end-to-end product work: I’ve built web platforms such as a medical social media app (React.js, Maplibre, Geoapify) and a medical assistance chatbot (Next.js, TypeScript, Flask, PyMongo) that connects user queries to an LLM server, and I continuously sharpen my fundamentals through mentoring and competitive programming.

Experience

Work history, roles, and key accomplishments

R&D Research Intern

Samsung Research Institute

May 2025 - Jul 2025 (2 months)

Deployed a transformer-based image captioning model (BLIP) delivering real-time performance with ≤2s latency and ≤200MB memory footprint on low-end mobile CPUs. Optimized BLIP and Gemma-1B models via selection, conversion, and quantization for efficient on-device inference, and fine-tuned YOLOv8 for domain-specific object detection using a custom annotated dataset.

Image Captioning Transformer Models (BLIP Model Quantization On Device Inference Optimization YOLOv8 Performance Tuning