Multimodal AI
Developing solutions to evaluate student responses across multiple modalities.
PhD Scholar · IIT Bombay
Developing data efficient AI alignment solutions.
I’m Pritam Sil, a PhD scholar in Computer Science at IIT Bombay. My research spans AI alignment, multimodal AI, SSL and AI in education , with a focus on building learning technologies that are useful, robust, and grounded in human judgment.
A little about me
Research grounded in real life educational contexts.
I’m a PhD researcher working on developing data-efficient AI alignment solutions for multimodal AI and I’m fortunate to be advised by -
- Prof. Ganesh Ramakrishnan · IIT Bombay
- Prof. Pawan Goyal · IIT Kharagpur
- Dr. Venkatapathy Subramanian · BharatGen
As part of BharatGen’s research initiatives, we collaborate with NGOs and other organizations to bring AI for Education research into practice. In the first year of my PhD, I had the privilege of being advised by the late Prof. Pushpak Bhattacharyya. I remain deeply grateful for his guidance, and his memory continues to shape my work.
My earlier MS Thesis includes developing an AI-assisted grading of handwritten answer sheets which was published at AIED 2024 and is currently deployed on the SAFE platform being used at IIT Bombay. Additionally, I have worked on research on multimodal human activity recognition.
“There’s nothing artificial about AI. It’s inspired by people, it’s created by people, and, most importantly, it impacts people.” - Fei Fei Li
What I work on
Research interests
LLM Evaluation & Alignment with Expert Judgment
Developing council-based verification of AI-generated educational questions, where SLMs act as judges aligned to expert decisions.
AI in Education
Designing AI teaching assistants, solutions for short-answer grading with feedback, and pronunciation assessment for language learners.
Noise-Robust Learning
Training methods that stay reliable under noisy labels and imperfect supervision, from classification to large language models.
Human-Centered AI
Building systems that assist teachers and learners rather than replace them, with explanations, feedback, and human in the loop solutions.
Speech & Language Learning
Developing semi-supervised techniques for automatic pronunciation assessment of L2 English speech.
Selected work
Publications
2026
How Effective Are VLMs in Assisting Humans in Inferring the Quality of Mental Models from Multimodal Short Answers?
P. Sil, D. Karnam, V. Venumuddala, and P. Bhattacharyya
2025
“I Understand Why I Got This Grade”: Automatic Short Answer Grading (ASAG) with Feedback
D. Aggarwal, P. Sil, B. Raman, and P. Bhattacharyya
Achieving Noise Robustness by Additive Normalization of Labels
M. Ghulyani*, P. Sil*, and A. Azad · *Equal contribution
2024
Can AI Assistance Aid in the Grading of Handwritten Answer Sheets?
P. Sil, P. Chaudhuri, and B. Raman
2021
MMHAR-EnsemNet: A Multi-Modal Human Activity Recognition Model
A. Das, P. Sil, P. K. Singh, V. Bhateja, and R. Sarkar
Finding Prediction of Interaction Between SARS-CoV-2 and Human Protein: A Data-Driven Approach
M. Ghosh, P. Sil, A. Roy, R. Fajriyah, and K. C. Mondal
Frequent Itemset Mining Using FP-tree: A CLA-based Approach and Its Extended Application in Biodiversity Data
M. Ghosh, A. Roy, P. Sil, and K. C. Mondal
Roles & positions
Experience
Research experience
-
Jul 2024 — now
Doctoral Researcher · IIT Bombay
Designing data efficient AI alignment solutions for multimodal answer evaluation · Advisors: Prof. Ganesh Ramakrishnan, Prof. Pawan Goyal, and Dr. Venkatapathy Subramanian
Research highlights
- Inferring student mental model quality from multimodal responses using VLMs.
- Automatic pronunciation evaluation for L2 English speech.
- Council-based verification of AI-generated educational Q&A.
- Multimodal short-answer grading with feedback; collaborating with NGOs and universities on AI for Education deployments.
-
Aug 2022 — Jun 2024
MS Thesis · IIT Bombay
Introduced AI-assisted grading to SAFE, developing an end-to-end pipeline for grading scanned handwritten answer sheets. Work published at AIED 2024. Advisors: Prof. Bhaskaran Raman and Prof. Parag Chaudhuri.
-
Aug — Dec 2021
R&D Project · Apache Spark Resource Usage Characterization
Studied relationships between system performance metrics and Spark parameters, and built a performance-monitoring tool. Advisor: Prof. Varsha Apte, IIT Bombay.
-
2018
Research Collaborations · Ohio University
- Co-developed an algorithm for predicting Pareto-optimal solutions in job-shop scheduling.
- Performed differential expression analysis for a thymoma gene-expression study.
Industry & other roles
-
May — Jul 2025
Research Intern · Fujitsu Research India
Researched on noise-robust training methods for large language models. Patent filed; manuscript to be submitted to TMLR.
-
2021 — present
Research Assistant · IIT Bombay
Maintain the SAFE iOS app and backend server, used in production at IIT Bombay. Advisor: Prof. Bhaskaran Raman.
-
2020
Data Science Summer Intern · IIT Guwahati
Developed an R Shiny app to forecast COVID-19 case counts at state and national levels. Advisor: Dr. Palash Ghosh, IIT Guwahati. More details here ↗
-
2018 — 2019
Technical Content Writer · GeeksForGeeks
Wrote technical articles on genetic algorithms, DNA cryptography, and Java. Browse all my articles ↗
Academic background
Education
2024 — now
PhD in Computer Science and Engineering
IIT Bombay
2024
MS by Research in Computer Science and Engineering
IIT Bombay
2021
BE in Information Technology
Jadavpur University
A small window into my travels and moments away from research.
@monday_backpacker
Snapshots from the road.
The latest posts from my Instagram feed. If the feed is unavailable, visit the profile directly.
Visit InstagramLet’s connect.
Interested in research, collaboration, or a conversation about AI for education?