Headline

Senior Machine Learning Scientist | Applied AI, Computer Vision & GenAI | Research to Production

About

I'm a machine learning scientist with over 4+ years of experience building applied AI systems, mainly in computer vision. My work usually starts with problems that are open-ended and not yet well defined, and my focus is on carrying them all the way through to something that works reliably in production rather than staying at the prototype stage. At Coachbuddy.ai I lead the AI and ML work behind racquet-sports analytics, from how the system perceives players and the shuttle to reconstructing rallies in 3D for replay and simulation. Earlier, at VuNet Systems, I worked on infrastructure and operations analytics, using machine learning to model large-scale transaction systems and make them easier to monitor and debug. I also come from a research background, with an M.E. in Machine Learning, a couple of filed patents, and published work in medical imaging. I'm drawn to problems that need both the research figured out and the engineering to make it run in the real world, and I care about solutions that are technically sound and actually useful.

Location

Bengaluru

AI stack

claude code

claude

codex

gemini

hugging face

obsidian

Work experience

Senior Machine Learning Scientist at Coachbuddy AI, May 2025 - Present · 1 yr 4 mos, Promoted to Senior Machine Learning Scientist with ownership of core AI systems for Coachbuddy’s racquet-sports analytics and intelligent training platform. • Own end-to-end ML development across perception, rally understanding, 2D-to-3D trajectory reconstruction, replay, simulation, and real-time deployment. • Build geometry-aware computer vision pipelines combining detection, tracking, pose estimation, camera calibration, homography, and constrained numerical optimization. • Optimize and deploy models for edge and on-device inference using ONNX, TensorRT, TFLite/LiteRT, quantization, benchmarking, and Android integration. • Contributed to filed patent applications related to automated racquet-rally replay and AI-based racquet-sport opponent simulation.

Machine Learning Scientist at Coachbuddy AI, Apr 2024 - Apr 2025 · 1 yr 1 mo, Worked on the foundational computer vision and ML systems for Coachbuddy’s badminton analytics platform. • Developed pipelines for court detection, player tracking, shuttle tracking, pose estimation, hit detection, and rally-event understanding. • Worked on camera geometry, homography, trajectory analysis, and early 2D-to-3D reconstruction experiments for sports analytics. • Built, evaluated, converted, and benchmarked ML models across PyTorch, TensorFlow, ONNX, TensorRT, and TFLite workflows.

Data Science Engineer at VuNet Systems, Jun 2022 - Apr 2024 · 1 yr 11 mos, Bengaluru, Karnataka, India, Worked on machine learning and analytics for large-scale infrastructure observability, capacity planning, root-cause analysis, and alert intelligence. • Researched and developed ML-based capacity planning methods for large UPI infrastructure systems to estimate server capacity required for incoming transaction traffic. • Worked on root-cause analysis of infrastructure and application metrics to identify bottlenecks across complex application journeys. • Developed alert reduction workflows by clustering related infrastructure alerts and using LLM-assisted summaries to provide meaningful operational insights. • Built improved Grafana dashboards and visualizations to help users understand system behaviour, traffic patterns, bottlenecks, and capacity trends more intuitively. • Implemented ML lifecycle-friendly Git management practices to improve experiment tracking, collaboration, and code organization across data science workflows.

Project Intern at Infineon Technologies, Jul 2021 - May 2022 · 11 mos, Bengaluru, Karnataka, India, Worked with Infineon’s IT department on an in-house data lake platform for large-scale file and data management. • Developed the complete frontend UI for the data lake platform using Vue.js. • Built and maintained reusable UI components and workflows for file management, search, and data operations. • Contributed to internal Python libraries used for API calls and platform integration.

ML/AI Development Intern at Paralaxiom Techologies Private Limited, Mar 2020 - Jun 2020 · 4 mos, India, Worked on backend development and automation workflows for computer vision and video analytics applications. • Built an automated training pipeline for TensorFlow-based object detection models using Python and shell scripting. • Developed Django APIs for Paralaxiom’s Video Analytics and Surveillance Toolkit (VAST). • Created a Cython-based packaging workflow to compile and package parts of the platform codebase for deployment.

Education

Master of Engineering - M.E., Machine Learning, Manipal School of Information Sciences

Bachelor of Technology, Electronics and Telecommunication, Shri Guru Gobind Singhji Institute of Engineering and Technology, Vishnupuri, Nanded

Public links

github: https://github.com/romillodaya