IIT Madras Zanzibar Campus

Pixel-based Fast Motion Estimation: Designed, developed, and implemented a novel Adaptive pixel-based direction-oriented fast motion estimation method for lossless video compression and extension towards Surveillance Video Coding.Medical Image Compression: Designed, developed, and implemented a groundbreaking context-based lossless image compression methodology specifically tailored for longitudinal (4D) medical images like CT and MRI.Deep learning-based Dose Prediction: Employed DL techniques to accurately predict dose and dose levels in radiotherapy planning, revolutionizing treatment planning processes with improved accuracy and efficiency.

Video Codec Algorithm Development: Designed and developed algorithms to enhance the compression efficiency of H264 (AVC)/H.265 (HEVC) encoder implementation in QC chipsets, collaborating with global teams within the organization.HDR10 and HEIF: Acquired proficiency in various video quality metrics and successfully integrated them for HDR10 and HEIF perceptual quality assessment in critical client applications.

Deep Tech Startup: During my time at Entrepreneur First, a renowned incubator for early-stage deep-tech founders, I gained valuable experience in team formation and idea creation. I actively participated in the ideation process and had the opportunity to explore a potential project involving the development of a lossless data compression framework.

Sakura exchange visitor under the framework of the Japan-Asia Youth Exchange program administered by the Japan Science and Technology Agency. Robotics and Deep Learning: Led the design, dataset preparation, and training using deep learning techniques for a robotic arm capable of playing rock-paper-scissors against a human. Programmed the robotic arm and collaborated with fellow graduates to seamlessly integrate the solution for a captivating live demonstration.

Skeleton Data Compression: Designed and developed an innovative framework for lossless compression of skeleton point data. Collaborated closely with esteemed faculties and research graduates at SJTU to ensure successful implementation.Efficient Motion Estimation for Surveillance Videos: Designed and developed a foreground-aware efficient motion search algorithm that is background, foreground, and boundary-aware, specifically designed for surveillance videos.

Conducting lab experiments and tutorials for undergraduate and postgraduate students.

Full Stack Development: Played a crucial role in developing and maintaining an end-to-end order management application for a prominent automotive client. Within this role, my responsibilities encompassed a wide range of programming tasks, including SQL query writing, analysis, development, implementation, modification, testing, and ongoing system maintenance.