CV

Education

• Ph.D in Artificial Intelligence, National University of Singapore, 2028 (expected)
• M.S. in Biotechnology, IIT Madras, 2021
• B.S. in Biotechnology, IIT Madras, 2021

Work experience

• (Feb 2024 - Present) Research Associate
  • National University of Singapore, Singapore
  • Supervisor: Professor Dianbo Liu
  • Research Area: VQVAE, Reinforcement Learning, Large Language Models, AI for Science
  • Research Project (Published):
    • Expected Return Causes Outcome-Level Mode Collapse in Reinforcement Learning and How to Fix It with Inverse Probability Scaling
• (August 2023 - December2023) Machine Learning Engineer
  • VAO Labs, California, USA
  • fine-tuning LLMs for advanced document understanding and data extraction across multiple business workflows.
  • Built and deployed scalable AI/ML pipelines that powered automated order entry and ERP data integration, significantly reducing manual processing time and errors.
  • Collaborated with engineering and product teams to improve generative AI accuracy for enterprise document workflows.
• (July 2022 - July 2023) Research Assistant
  • IIT Madras, India
  • Supervisor: Professor Srinivas Chakravarthy
  • Research Project (published):
    • Brain-Inspired Attention Model for Object Counting
  • Project (Patent Granted):
    • Designing and developing an AI based mobile application for Alziehmer patients to help them with daily tasks.
    • Designing and developing an AI based mobile application for Parkinson’s patients to help them diagnose and monitor their symptoms using their walking patterns.
• (June 2021 - December 2021) Software Engineer
  • IIFL Wealth Management, Mumbai, India
  • Responsibilities include:
    • Updates and improvements to the trading platform
    • Testing and quality assurance
    • Team collaboration

Skills

• Core Machine Learning & Representation Learning
  • Supervised, Unsupervised & Semi-Supervised Learning
  • Discrete Representation Learning & Vector Quantization (e.g., VQ-VAE)
  • Latent Variable Models & Generative Modeling
  • Feature Learning and Embedding Optimization
• Large Language Models (LLMs)
  • LLM Fine-Tuning & Adaptation
  • Prompting Strategies and Instruction Tuning
  • Continual & Transfer Learning for LLMs
  • Reasoning-Enhanced LLM Training & Evaluation
• Reinforcement Learning for LLMs & Reasoning
  • Reinforcement Learning (RL) methods for enhancing LLM reasoning
  • Reward design including verifiable and structured rewards
  • Policy Optimization techniques for sequence models
  • Training strategies such as actor-critic, critic-free, and hybrid RL
  • Algorithms for reasoning tasks (e.g., reasoning correctness, planning)
• Algorithms & Advanced Methods
  • RL from Demonstrations / Reinforcement Learning from Human Feedback (RLHF)
  • Reinforcement Learning with structured & dynamic sampling
  • Integrating RL with supervised-learning baselines
  • Techniques for reasoning, planning, reflection & self-correction in LLMs
  • Research on verifiable reward functions & evaluation metrics
• Systems & Tools
  • Languages: Python (primary research language), SQL
  • ML Frameworks: PyTorch, TensorFlow, Hugging Face Transformers
  • RL Libraries: Stable Baselines, RLlib, custom implementations
  • Experiment Management: Weights & Biases, MLflow
  • Compute: GPU/TPU training workflows, distributed training

Publications

Outcome level mode collapse in reinforcement learning

Physical Plausibility Failure Modes with Matryoshka Transcoders

Brain-Inspired Attention Model for Object Counting

Cocurriculars

• Trying to understand the meaning of life and 42
• Integrating modern lifestyle with different existential philosophies of the eastern and western world
• Practising YOGA as a way of life
• Certified Yoga Instructor from Global School of Yoga Alliance to teach and discuss yoga philosophy and practice.