Abhinit Negi

AI/ML Engineer @ Visa | Computer Vision, NLP | Ex-Mercedes-Benz R&D

Bengaluru, Karnataka, India

About

AI/ML Engineer with 3+ years of experience building and deploying production machine learning systems at Visa Inc. and Mercedes-Benz R&D. I specialize in: - Generative AI & LLMs: Designed and deployed RAG-based chatbots using large language models, impacting 700+ users - Computer Vision & 3D Perception: Built deep learning models for 3D obstacle detection and panoptic segmentation on point-cloud data - NLP & Information Retrieval: Developed NLP pipelines saving 300+ hours/week of manual work - MLOps & Cloud: End-to-end ML pipeline development on AWS, Azure, GCP with Docker and CI/CD Core skills: Python, PyTorch, TensorFlow, LangChain, Hugging Face, Scikit-learn, SQL, Spark, Docker, Kubernetes, AWS SageMaker, Azure ML, MLflow Open to opportunities in AI/ML Engineering, Data Science, and Applied Research.

Experience

  • AI Engineer/ Data Scientist at Visa
    Jul 2025 - Present · 1 yr 1 mo

  • Machine Learning Engineer at Mercedes-Benz Research and Development India
    Aug 2023 - Jul 2025 · 2 yrs

    – Developed and deployed Retrieval-Augmented Generation(RAG) based chat bots based on LLMs to make the internal document retrieval seamless thereby impacting 700+ employees. – Created an evaluation pipeline to test the response of the Retrieval-Augmented Generation(RAG) based chat-bot. – Performed rigorous code reviews, improved code quality by 30 % and ensured maintainability, scalability, and readability, resulting in faster on-boarding and a 20% reduction in debugging time. – Collaborated with cross-functional teams to leverage NLP based models for effortless fetching of quality control documents enhancing workforce efficiency by saving 300+ hours of manual work per week. - Developed, enhanced and deployed deep learning models for 3D obstacle detection on point-cloud data, the lead to increase in system accuracy by 11 percent. – Designed and optimized machine learning models for panoptic segmentation of in-house point-cloud data achieving mean average precision(mAP) improvement of 7 percent.

  • Research Assistant at National Institute of Technology Karnataka
    May 2022 - May 2023 · 1 yr 1 mo