Hello, I'm
Kushal Borkar.

I am a Research-focused ML Engineer.

“The more I learn, the more I learn how little I know - and that certainly makes me more curious to learn more.”

Nice to meet you!

Computer VisionMedical ImagingApplied Deep Learning

01

About Me

I’m a research-focused Machine Learning Engineer working on multimodal models and geometry-aware deep learning for long-form video understanding. My work has centered on building systems that move beyond simple detection toward predictive, self-supervised reasoning, combining Vision-Language-Action ideas with JEPA/MAE-style learning to capture both pixel-level detail and high-level temporal dynamics. I’ve also worked on universal ground-plane calibration for sub-5 inch localization across arbitrary sports and camera angles, and on interactive medical image segmentation with continual test-time adaptation for robust performance in out-of-distribution clinical settings.

I hold an MS by Research in Computer Science from IIIT-Hyderabad, where I was fortunate to work under Prof. C. V. Jawahar and Prof. Chetan Arora, where my research focused on interactive segmentation frameworks that leverage Continual Test-Time Training to achieve state-of-the-art robustness in out-of-distribution clinical settings. This shaped how I think about research rigor, problem selection, and deploying ideas that actually survive real-world constraints. I also hold a B.Tech (Hons.) in Electronics and Communication Engineering from IIIT SriCity.

Research & Engineering Focus:

  • Video & Action Models: Developing vision-language systems that understand video.
  • Continual Test-time Domain Adaptation: Building self-supervised models that adapt seamlessly to new environments upon deployment.
  • Deployable AI: Bridging the gap between theory and engineering to create robust, production-ready systems.
  • Domain Applications: Applying advanced AI to solve complex challenges in clinical healthcare and sports analytics.

Every so often, when not occupied with work or research, I set aside some time for watching anime, tuning in to podcasts, and planting and taking care of my plants.

ICVGIP Best Paper Runner Up Award recipientInvited to present – India AI SummitPublications in Elsevier, Frontiers, and ACM
Kushal Borkar

02

Latest News

2026
  • Invited to present at the India AI Summit, Delhi — representing research on continual learning and interactive segmentation.

2025
  • Successfully defended my Master's By Research thesis — Continual Learning in Interactive Medical Image Segmentation.

2024
  • Research paper awarded Best Paper Runner-Up at ICVGIP 2024.

  • Joined Thelios.AI as a Computer Vision Engineer.

2022
  • Began Master's research at IIIT Hyderabad under the guidance of Prof. C. V. Jawahar and Prof. Chetan Arora.

03

Experience

Computer Vision Engineer

Thelios.AI

  • Develop research-focused computer vision systems for long-form sports video understanding and geometry-aware localization.
  • Contribute to a dual-path MAE+JEPA video representation model for tubelet reconstruction and latent prediction, designed to capture both spatial detail and temporal dynamics.
  • Work on generalized homography pipelines for cross-sport ground-plane calibration under real-world deployment constraints.
  • Impact: Expanded research capability within the team by owning technically difficult modeling work at the intersection of self-supervised video learning and deployable sports analytics.
Jul 2025 – Present

Pune, India

Computer Vision Engineer, Contract

Level.AI

  • Built real-time computer vision pipelines for screen activity compliance monitoring using object detection and OCR.
  • Optimized video processing by reducing redundant frames through SSIM-based sampling while preserving key application state changes.
  • Improved the efficiency and scalability of production monitoring workflows by making real-time compliance tracking practical under deployment constraints.
Dec 2023 – Jun 2024

Remote

Research Fellow

International Institute of Information Technology, Hyderabad

Conducted research in interactive medical image segmentation and continual test-time adaptation under Prof. C. V. Jawahar and Prof. Chetan Arora. Developed robust segmentation frameworks for out-of-distribution clinical settings and contributed to annotation tooling for medical volume segmentation.

Jul 2022 – Dec 2024

Hyderabad, India

Research Fellow

International Institute of Information Technology, Hyderabad

Developed AI-based solutions in healthcare under the Intel Applied AI in Healthcare initiative, focusing on computer vision and deep learning for clinical imaging problems.

Jul 2020 – Jun 2021

Hyderabad, India

Data Science & Data Analyst Intern

Varidus

Improved accessibility of multi-source data for an AR-VR product by automating scraping across five data sources. Built regression-based analytics support for product decision-making.

Jan 2020 – Jul 2020

Singapore (Remote)

Data Science Intern

Masho

Designed a content-based filtering recommendation system using word2vec and similarity metrics to generate personalized product recommendations.

Sep 2019 – Dec 2019

Bangalore, India

04

Education

International Institute of Information Technology, Hyderabad

Master of Science (MS) by Research · Computer Science and Engineering

Research under Prof. C V Jawahar at the Human and AI Lab (HAI), focusing on computer vision, neuroimaging, and applied AI in healthcare as part of the Intel Applied AI Center (INAI) initiative.

Jan 2021 – Present

Hyderabad, India

Indian Institute of Information Technology, Sri City

Bachelor of Technology (Hons.) · Electronics and Communication Engineering

Aug 2015 – May 2019

Sri City, Andhra Pradesh, India

05

Publications

Recent Research Papers

2024
No Prompting Frozen Foundation Models: Interactive Medical Volume Segmentation using Continual Test Time Adaptation of Compact Models

Kushal Borkar, Abhilaksh Singh Reen, CV Jawahar, Chetan Arora

ICVGIP 2024, ACMBest Paper Runner UP
2023
PREHOST: Host prediction of coronaviridae family using machine learning

Anusha Chaturvedi, Kushal Borkar, U Deva Priyakumar, Vinod P.K.

Heliyon, CellPress
2022
Ayu – Characterization of Healthy Ageing from Neuroimaging Data with Deep Learning and rsfMRI

Kushal Borkar, Anusha Chaturvedi, Vinod P.K., Raju Surampudi Bapi

Frontiers in Computational Neuroscience
2020
2018
Video Dehazing Using LMNN with Respect to Augmented MRF

Kushal Borkar, Snehasis Mukherjee

ICVGIP 2018, ACM
Single Image Dehazing Based on Generic Regularity

Kushal Borkar, Snehasis Mukherjee

arXiv:1808.08610

06

Projects

A selection of research and engineering work.

Interactive Machine Learning
Computer VisionActive LearningSegmentationHuman-in-the-Loop

Interactive Machine Learning

Outlining an organ on a CT or MRI scan means tracing it on a hundred slices or more, one at a time. We built a way to make that far less painful.

Read more
Ayu: Characterization of Healthy Ageing from Neuroimaging
Medical ImagingrsfMRIDeep LearningNeuroscience

Ayu: Characterization of Healthy Ageing from Neuroimaging

Ayu estimates a person's age from a resting-state fMRI scan, then uses the model's attention to show which brain connections change as we get older.

Read more
Covid-19 Cough Detector
Audio MLSignal ProcessingCNNHealthcare AI

Covid-19 Cough Detector

A COVID-19 screening tool that listens to a short cough recording and predicts whether it's positive. Our team placed 7th in the Microsoft Azure AI Challenge.

Read more

08

Get In Touch

I am open to research collaborations, academic discussions, and industry opportunities in computer vision and applied AI systems. Feel free to reach out.

Let's Talk