Tarun Chaubey

About Tarun Chaubey

I\'m Tarun Chaubey, a B.Tech. Computer Science graduate who has been working as an Associate Data Scientist at Celebal Technology in Jaipur since November 2022. In my work, I am responsible for analyzing client objectives, identifying appropriate tech stacks, and creating end-to-end pipelines for data comprehension, model development, and production deployment.Over the years, I\'ve handled the development of several SAAS platforms. In the Computer Vision domain, I came up with a framework for detecting anomalies on overhead line towers with drone-captured images, as well as one for solar panel anomaly detection via real-time process control system (RTPS) frames. These solutions are seamlessly linked with the clients\' systems. At GenAI, I created EKA and post-call analysis tools that allow customers to extract essential insights from audio call recordings and documents, such as key points, phrases, Named Entity Recognition (NER), sentiment analysis, and Customer Satisfaction (CSAT) scores.My experience includes Python, Machine Learning, Deep Learning, Natural Language Processing, and Generative AI tools such as OpenAI and Gemini. I use Power BI and Tableau for visualization, and MLOps tools like Docker, Kubernetes, and Git on the Azure Cloud platform for deployment. In addition, I work directly with the Cloud & Infrastructure team to resolve network and security with solution with public or client environments.

Experience

1. Ai Engineer

   Ascentt

   Jul 2024 — Present · Indore, IN

   Developed AI applications (EKA Chatbot, Resume Bot, Vehicle Search) using OpenAI models & open-source embeddings, and Open Search for vector database management.Fine-tuned models to extract entities from documents for improved data processing.Optimized model throughput with quantization techniques (GGUF, ONNX, GGML).Built scalable backend with socket layer for real-time responses.Deployed and optimized Computer Vision models (classification, detection, and segmentation) for inference using ONNX, Open VINO, and Tensor RT across CPU and GPU.Developed a semi-automatic annotation tool using SAM for auto image labeling, integrated with Label Studio, enabling faster annotation and optimized inference performance.

Education

• St. Xavier's School Ballia

  Senior Secondary

  2016 — 2018

• Dr. A.P.J. Abdul Kalam Technical University

  Bachelor of Technology - BTech, Computer Science

  2018 — 2022