TUHIN ACHARJEE

Computer Vision Engineer with 3+ years of experience, currently working at Adagrad AI, where I design and deploy advanced vision-based systems for real-world applications, including people tracking, vehicle analytics, and ANPR (Automatic Number Plate Recognition).

Previously, I served as Tech Lead at Rosa Technology, leading the development of a computer vision-based rebar detection system that improved testing efficiency by 60%, along with building high-performance, low-latency microservices. I also developed a RAG-based AI chatbot integrated with robotics and mobile platforms, created a Python library for precise IMU sensor calibration, and built an LLM evaluation and observability framework that reduced hallucinations and improved response quality.

In my current role, I extensively work with C++, GStreamer, NVIDIA DeepStream, TensorRT, and vLLM to build optimized, real-time video analytics pipelines and scalable AI systems.

My broader expertise includes Python, FastAPI, HuggingFace, FAISS, Streamlit, NLP, and scalable system design. I’m passionate about building intelligent, AI-driven systems that create real-world impact across domains like smart surveillance, robotics, and automation, and I thrive in fast-paced, innovation-driven environments.