Shourov Joarder
Adjunct Lecturer,
CSE, BRAC University
Machine Learning Engineer,
ACI Limited
With the recent advancements in state-of-the-art multimodal AI, I am passionate about building real-world systems that are not only accurate but also fundamentally reliable and fair with explainable reasoning capabilities.
My work focuses on the critical challenges of fairness, generalization, and reasoning and test time training of multimodal models, particularly in high-stakes applications like healthcare, autonomous vehicles etc. I have a background in Electrical and Electronic Engineering and professional experience building and deploying computer vision and multimodal models.
About Me
I have broad interests in multimodal AI, NLP (LLM, VLM), and trustworthy machine learning test time training and autonomous vehicles. My hands-on research and professional experience have shaped my direction towards solving real-world problems with innovative and robust AI systems. I am particularly interested in:
- Addressing and mitigating bias in Vision-Language Models (VLMs) to ensure fairness and prevent the amplification of biases in medical and general VQA tasks.
- Optimization and Test Time Training of foundation models.
- Improving the general video understanding and reasoning capabilities of state-of-the-art foundation models.
- Developing robust deep learning methods for niche domains like medical imaging, such as my work on a novel two-stage unsupervised deep learning architecture for Ultrasound Strain Elastography (USE).
