Embedded Files
LCS2 is research group led by Dr. Tanmoy Chakraborty at IIT Delhi. Research at LCS2 has been focusing on advancing LLMs grounded in real-world human behavior, with four tightly integrated directions: (i) efficient and controllable LLMs, including parameter-efficient fine-tuning, knowledge distillation, pruning, KV-cache compression, context compression, and long-context or multilingual reasoning; (ii) behavior-aware and human-centric modeling, where language technologies are combined with insights from psychology and social science to study misinformation, counter-speech, cultural literacy, and the cognitive dynamics of online discourse; (iii) interpretability and diagnostic analysis, where we understand emerging, multimodal and multilingual capabilities of LLMs, and (iv) AI for high-stakes social applications, particularly mental health, encompassing conversational agents, counseling summarization, multimodal risk detection, and privacy-preserving learning frameworks. Across these themes, the lab emphasizes principled evaluation β probing generalization, fidelity, reasoning, and alignment, while developing multimodal, multilingual, and personalization-aware systems that are both scalable and deployable in real-world settings.
Want to work with us?
ποΈ To apply for MSR / PhD position, click here.
πTo apply for the Research Assistant (RA) / Post Doctoral position, drop a mail to chak.tanmoy.iit@gmail.com
News
May 1, 2026: Two papers on Concept Learning, Model Interpretability (ICML Spotlight: Top 2.2% papers), accepted in ICML'26.
April 7, 2026: Six papers (Interpretability, Personalization, Agentic SVLMs, Benchmarking), accepted in ACL'26.
February 8, 2026: Our paper on Mental Health Screening App got accepted on JMIR Formative Research!
January 28, 2026: Prof. Tanmoy Chakraborty received OpenAI Mental Health Grant Award. Congratulations!
January 27, 2026: Our paper on Memory for Personalized Summarization accepted in ICLRβ26.
January 13, 2026: Three papers (taxonomy expansion, LLM personalization, mental health), accepted in The WebConf'26.
January 04, 2026: Two papers (long-context multilingual reasoning, spatial symbolic reasoning), accepted in EACLβ26 (1 in Main, 1 in Findings)
December 16, 2025: Prof. Tanmoy Chakraborty received the NIVIDIA Academic Research Grant! Congratulations! π
December 03, 2025: Prof. Tanmoy Chakraborty received the INSA Young Associates! Congratulations! π
November 22, 2025: Our paper on scaling laws, accepted in TMLR.
November 8, 2025: Prof. Tanmoy Chakraborty Received the ACM India Outstanding Contribution to Computing Education (OCCE) 2025 Award.
November 08, 2025: Our paper on quantum-inspired taxonomy generation got accepted to AAAI'26.
Featured Highlight
Parmanu builds efficient AI architectures and compression to make powerful language models practical, affordable, and deployable on everyday hardware.
Manashi builds human-centered AI for mental healthcare, combining clinical insight, multilingual NLP, and ethical, real-world deployment to support clinicians.
Google Sites
Report abuse