Tajamul Ashraf
email

CV | Github | Linkedln
Google Scholar | Blog

I am currently a Research Associate II at MBZUAI in the Intelligent Visual Analytics Lab, primarily advised by Prof. Salman Khan and Prof. Rao Anwer. Previously, I was a Research Intern at Microsoft Research India in Bengaluru. Recently, I completed my Master's degree in Computer Science from the Indian Institute of Technology Delhi under the guidance of Prof. Chetan Arora. My academic journey began with Bachelor's degree in Information Technology from the National Institute of Technology, Srinagar.

Outside of work, I enjoy playing badminton and swimming quite often. I'm also passionate about making a positive impact on the society which led me to initiate Ralith Milith, an anti-drug society in Kashmir.

"What is now proved was once only imagined." -William Blake.

News

Research

My research interests lie at the intersection of multimodal systems in vision and language domains, with a focus on enhancing reasoning and decision-making capabilities. I am particularly interested in addressing the challenges faced by large multi-modal models across discriminative, generative, and perceptual understanding tasks, especially in Out-Of-Distribution (OOD) and federated scenarios. My work explores domain adaptation, reinforcement learning, causal reasoning, and knowledge distillation techniques to improve the robustness and generalization of AI models in complex, real-world environments such as healthcare and robotics.

Publications

  • Selected
  • All
Profile Picture

LLM Post-Training: A Deep Dive into Reasoning Large Language Models
Komal Kumar*, Tajamul Ashraf*, Omkar Thawakar, Rao Muhammad Anwer, Hisham Cholakkal, Mubarak Shah, Ming-Hsuan Yang, Phillip H.S. Torr, Fahad Shahbaz Khan, Salman Khan
Arxiv 2025

webpage | abstract | code

Profile Picture

D-MASTER: Mask Annealed Transformer for Unsupervised Domain Adaptation in Breast Cancer Detection from Mammograms
Tajamul Ashraf, Krithika Rangarajan, Mohit Gambhir, Richa Gabha, Chetan Arora
MICCAI 2024 [Poster]

webpage | abstract | presentation | code

Profile Picture

TransFED: A way to epitomize Transformer based Focal Modulation using Federated Learning
Tajamul Ashraf, Fuzayil Mir, Iqra Altaf Gillani
WACV 2024 [Poster]

webpage | abstract | presentation | code

Profile Picture

HF-Fed: Hierarchical Based Customized Federated Learning Framework for X-Ray Imaging
Tajamul Ashraf, Tisha Madame
MICCAI DeepBreath 2024 [Oral] Best Student Paper Award!

webpage | abstract | presentation | code

Profile Picture

LLM Post-Training: A Deep Dive into Reasoning Large Language Models
Komal Kumar*, Tajamul Ashraf*, Omkar Thawakar, Rao Muhammad Anwer, Hisham Cholakkal, Mubarak Shah, Ming-Hsuan Yang, Phillip H.S. Torr, Fahad Shahbaz Khan, Salman Khan
Arxiv 2025

webpage | abstract | code

Profile Picture

Enhancing Climate Change Understanding: A Novel Deep Learning Framework with the Climate Change Parameter Model.
Tajamul Ashraf, Janibul Bashir
MoSICom 2024

webpage | abstract | code

Profile Picture

Phase-Informed Tool Segmentation for Manual Small-Incision Cataract Surgery
Bhuvan Sachdeva, Naren Akash, Tajamul Ashraf, Simon Muller, Thomas Schultz, Maximilian WM Wintergerst, Niharika Singri Prasad, Kaushik Murali, Mohit Jain
Arxiv 2024

webpage | abstract | presentation | code

Profile Picture

FATE: Focal-modulated Attention Encoder for temperature prediction
Tajamul Ashraf, Janibul Bashir
ArXiv 2024

webpage | abstract | code

Profile Picture

D-MASTER: Mask Annealed Transformer for Unsupervised Domain Adaptation in Breast Cancer Detection from Mammograms
Tajamul Ashraf, Krithika Rangarajan, Mohit Gambhir, Richa Gabha, Chetan Arora
MICCAI 2024 [Poster]

webpage | abstract | presentation | code

Profile Picture

HF-Fed: Hierarchical Based Customized Federated Learning Framework for X-Ray Imaging
Tajamul Ashraf, Tisha Madame
MICCAI DeepBreath 2024 [Oral] Best Student Paper Award!

webpage | abstract | presentation | code

Profile Picture

TransFED: A way to epitomize Transformer based Focal Modulation using Federated Learning
Tajamul Ashraf, Fuzayil Mir, Iqra Altaf Gillani
WACV 2024 [Poster]

webpage | abstract | presentation | code

Profile Picture

PoseWatch: Advancing Real Time Human Pose Tracking and Juxtaposition with Deep Learning
Tajamul Ashraf, Balaji Prabu BV, and Omkar SN
CVIP 2023

abstract | presentation

Profile Picture

Climate Change Parameter Dataset (CCPD): A Benchmark Dataset for Climate Change parameters in Jammu and Kashmir
Tajamul Ashraf, Janibul Bashir
ICDSA 2023

abstract | presentation

Profile Picture

An Integral Computer Vision System for Apple Detection, Classification, and Semantic Segmentation,
Tajamul Ashraf, Naiyer Abbas, Mohammad Haseeb, Nadeem Yousuf, Janibul Bashir
ICMV 2022

abstract | presentation | code

Timeline

  • MBZUAI Jan 2025 - Present
    Research Associate II
    Advisor(s): Prof. Salman Khan, Prof. Rao Anwar
  • MICROSOFT RESEARCH July 2024 - Dec 2024
    Research Intern
    Advisor(s): Dr. Mohit Jain, Dr. Kalika Bali
  • IIT DELHI Aug 2022 - June 2024
    Master's of Science (Research)
    Advisor(s): Prof. Chetan Arora
  • LEENA AI May 2022 - July 2022
    Software Developer
    Product, Engineering and ML team
  • ARTPARK IISc Jan 2022 - May 2022
    Computer Vision Researcher
    Mentor: Prof. Raghu Krishnapuram
  • IIT KHARAGPUR Nov 2019 - Mar 2020
    Research Intern
    Advisor(s): Prof. Pabitra Mitra
  • NIT SRINAGAR Aug 2018 - May 2022
    B-Tech in Information Technology)
    Advisor(s): Prof. Janibul Bashir
  • TYNDALE BISCOE SCHOOL 2016 - 2018
    Higer Secondary
    Class X with 92.6%
    Class XII with 91.8%

Languages I can connect in


English


Kashmiri


Urdu


Hindi


Arabic


Persian


Community

    In my academic journey, I transitioned into a research-focused lifestyle, driven by a deep curiosity to explore Computer Vision and Medical Imaging. My path has been dynamic, with a domain shift from IoT to robotics, reflecting a continuous pursuit of learning and innovation. I am an advocate for open-source contributions, as they foster collective growth and help both educators and learners critically engage with the wealth of information available online. The guidance and support of my academic mentors and peers have been pivotal in shaping my early career, and their influence continues to inspire me.
    Building on this foundation, I am actively mentoring undergraduate and master’s students in computer vision, and I look forward to supporting more learners in the future. I particularly encourage students with diverse backgrounds or unconventional academic paths, similar to my own, to connect and explore opportunities for growth and research.
    If you are interested, send an introductory email that includes:
    - A brief introduction about yourself.
    - Your academic background and areas of interest.
    - Your CV (optional but preferred).