Arka Mukherjee

arka.mukherjee078@gmail.com

Bhubaneswar, India

Hi, I’m Arka Mukherjee, a KIIT CS junior passionate about research in multimodal LLMs, evaluation, and reasoning with a focus on human-computer and agent-computer interaction. Currently, I am a funded Research Fellow at IIT Bhubaneswar with Dr. Shreya Ghosh. Previously, I spent a summer at the VLED lab, IIT Ropar, and contributed to AI security research at RespAI Lab, KIIT.

Beyond research, I engage with the tech community as a tech journalist and YouTuber, where I share insights on GPUs and emerging trends in AI.

news

Nov 11, 2025	mmJEE-Eval is now live! Check it out at mmjee-eval.github.io.
Nov 01, 2025	Team Blackwell ranked #6/59 (top 10th percentile) in the NeurIPS 2025 DCVLR data curation challenge!
Oct 19, 2025	Attended ICCV 2025 in Honolulu, Hawaii to present my social AI research.
Oct 17, 2025	Work on alternative eval metrics beyond accuracy for math and science reasoning accepted to the NeurIPS 2025 MATH-AI workshop!
Jul 12, 2025	Cultural competence VLM paper accepted to ICCV 2025 ASI workshop as an Oral presentation!

selected publications

ICCV
Toward Socially Aware Vision-Language Models: Evaluating Cultural Competence Through Multimodal Story Generation

Arka Mukherjee and Shreya Ghosh

In Proceedings of the IEEE/CVF International Conference on Computer Vision Workshops (ICCVW), Oct 2025

Oral Abs Bib HTML PDF

Oral Presentation

We present the first systematic evaluation framework for assessing cultural competence in Vision-Language Models (VLMs) through multimodal story generation, analyzing five contemporary VLMs with novel evaluation metrics.
@inproceedings{mukherjee2025cultural, title = {Toward Socially Aware Vision-Language Models: Evaluating Cultural Competence Through Multimodal Story Generation}, author = {Mukherjee, Arka and Ghosh, Shreya}, booktitle = {Proceedings of the IEEE/CVF International Conference on Computer Vision Workshops (ICCVW)}, year = {2025}, month = oct, }
AACL
mmJEE-Eval: A Bilingual Multimodal Benchmark for Evaluating Scientific Reasoning in Vision-Language Models

Arka Mukherjee and Shreya Ghosh

Findings of IJCNLP-AACL 2025, Nov 2025

Abs arXiv Bib HTML PDF Website

We introduce mmJEE-Eval, a bilingual multimodal benchmark comprising 1,460 STEM problems evaluating 17 VLMs. Models detect 53% of errors but correct only 3.5%, exposing metacognitive gaps between open and closed models.
@article{mukherjee2025mmjee, title = {mmJEE-Eval: A Bilingual Multimodal Benchmark for Evaluating Scientific Reasoning in Vision-Language Models}, author = {Mukherjee, Arka and Ghosh, Shreya}, journal = {Findings of IJCNLP-AACL 2025}, year = {2025}, month = nov, }