Deep Mathukiya

I’m an AI/ML Engineer with hands-on experience building and deploying production-oriented AI systems across Generative AI, RAG, Computer Vision, Edge AI, and Machine Learning. I have a strong interest in developing practical AI solutions that move beyond experimentation and can be deployed efficiently in real-world environments.

My recent work includes building LLM/RAG systems and developing an MCP-based web search tool that enables AI agents to interact with and retrieve information from the internet at significantly lower cost than conventional approaches. I’ve also worked extensively on Computer Vision and Edge AI, including a real-time person recognition system using YOLO and InsightFace, optimized for NVIDIA T4 and Rockchip NPU hardware.

I have built and contributed to several end-to-end AI projects, including a Gujarati Sign Language-to-Speech system deployed on NVIDIA Jetson hardware, a flood prediction and mapping platform using machine learning and geospatial data, an AI-powered agricultural disease detection system, and a PDF information extraction and RAG chatbot using NVIDIA Nemotron and vector databases.

I’ve participated in and won multiple competitive technology events, including the NVIDIA Nemotron 3 Super Contest, and secured 3rd place at the FloodAI Hackathon organized by E-Cell IIT Gandhinagar. My research work in Voice-to-Sign and Sign-to-Speech translation has also been accepted for publication through Springer Nature.

My technical stack includes Python, PyTorch, TensorFlow, Scikit-learn, OpenCV, YOLO, Hugging Face, LLMs, RAG, vector databases, FastAPI/Flask, React, Docker, MongoDB, MySQL, and cloud/edge deployment technologies.

I’m particularly interested in opportunities where I can work on AI/ML engineering, Generative AI, LLM/RAG systems, Computer Vision, Edge AI, AI agents, or applied research. I enjoy working on challenging problems, building systems from the ground up, optimizing models for real-world constraints, and collaborating with teams that are pushing AI into practical applications.