Het Shah profile photo 1 on Tal

Het Shah

Researcher

Delhi

2+ years of work experience

Wadhwani AI
Jan 20266 mosPresent

Delhi, India

- LLM-Based Evaluation Metrics: Designing human interpretable metrics to effectively evaluate generations using language models. - Enhancing RAG QA systems and Exploring Conformal Predictions with RAGs.
Artificial Intelligence Institute of South Carolina
Jan 20251 yrDec 2025

Columbia, South Carolina, United States

Advisor: Dr. Amit Sheth - Worked on Neurosymbolic AI to integrate AI with MTSS K-12 education framework. - Engineered a multimodal content pipeline utilizing LLaVA-mini and OCR to analyze visual and transcript data, autonomously generating timestamped, difficulty-graded assessment questions from raw lecture videos. - Developed a personalized learning engine that uses hybrid reinforcement learning for real-time difficulty adaptation and a Llama-3.2-based RAG system to generate context-aware, user-tailored summaries. - Co-authored a research paper ‘‘PAL: Personal Adaptive Learner,’’ accepted for publication at AAAI 2026.
AI-NLP-ML Research Lab IIT Patna
May 20255 mosSep 2025

Patna, Bihar, India

Advisor: Dr. Sriparna Saha Implemented a Group Relative Policy Optimization (GRPO) algorithm integrated with knowledge graphs and vector databases in PyTorch for Indian cultural question-answering.
FLaMe.nlp Research Lab
Jan 20241 yr 6 mosJun 2025

New Delhi, Delhi, India

Advisor: Dr. Md. Shad Akhtar - Improved LLM faithfulness evaluation in mental health conversations by designing a novel fused-attention architecture in PyTorch, integrating a RAG module with a knowledge-distilled CoT model. - Co-primary author in a research paper accepted at ACL 2026 Main Conference for Oral presentation.

Education

Indraprastha Institute of Information Technology, Delhi

Bachelor of Technology - BTech, Computer Science and Engineering

Indraprastha Institute of Information Technology, Delhi

20224 yrs2026
Indian School Wadi Kabir (CBSE)

High school diploma

Indian School Wadi Kabir (CBSE)

20202 yrs2022

Join Het and 10k+ other professionals on

Tal Boss

Tal is the app where bosses meet engineers & builders directly, without middlemen

Download Tal Boss

Het Shah on Tal

Headline

Researcher

About

Hi, I'm Het, a recent Computer Science graduate from IIIT Delhi focused on Deep Learning, Natural Language Processing, and Reinforcement Learning. I love turning ideas into working systems and collaborating on projects that solve real problems. Feel free to contact me: het22213@iiitd.ac.in

Location

Delhi

Work experience

Machine Learning Intern at Wadhwani AI, Jan 2026 - Present · 6 mos, Delhi, India, - LLM-Based Evaluation Metrics: Designing human interpretable metrics to effectively evaluate generations using language models. - Enhancing RAG QA systems and Exploring Conformal Predictions with RAGs.

Research Intern at Artificial Intelligence Institute of South Carolina, Jan 2025 - Dec 2025 · 1 yr, Columbia, South Carolina, United States, Advisor: Dr. Amit Sheth - Worked on Neurosymbolic AI to integrate AI with MTSS K-12 education framework. - Engineered a multimodal content pipeline utilizing LLaVA-mini and OCR to analyze visual and transcript data, autonomously generating timestamped, difficulty-graded assessment questions from raw lecture videos. - Developed a personalized learning engine that uses hybrid reinforcement learning for real-time difficulty adaptation and a Llama-3.2-based RAG system to generate context-aware, user-tailored summaries. - Co-authored a research paper ‘‘PAL: Personal Adaptive Learner,’’ accepted for publication at AAAI 2026.

Research Intern at AI-NLP-ML Research Lab IIT Patna, May 2025 - Sep 2025 · 5 mos, Patna, Bihar, India, Advisor: Dr. Sriparna Saha Implemented a Group Relative Policy Optimization (GRPO) algorithm integrated with knowledge graphs and vector databases in PyTorch for Indian cultural question-answering.

Undergraduate Researcher at FLaMe.nlp Research Lab, Jan 2024 - Jun 2025 · 1 yr 6 mos, New Delhi, Delhi, India, Advisor: Dr. Md. Shad Akhtar - Improved LLM faithfulness evaluation in mental health conversations by designing a novel fused-attention architecture in PyTorch, integrating a RAG module with a knowledge-distilled CoT model. - Co-primary author in a research paper accepted at ACL 2026 Main Conference for Oral presentation.

Education

Bachelor of Technology - BTech, Computer Science and Engineering, Indraprastha Institute of Information Technology, Delhi

High school diploma, Indian School Wadi Kabir (CBSE)

Het Shah - Machine Learning Intern in Delhi | Tal