Rajesh Debnath
CS PhD Student at North Carolina State University
Email: rajeshnath18snb@gmail.com | Phone: +1 (984) 789-6412
Location: Raleigh, NC
Summary
CS PhD @ NCSU focused on offline & hierarchical RL for Intelligent Tutoring Systems, robust evaluation (DR/WDR), and sequence modeling. Built end-to-end ML systems (PyTorch, d3rlpy); deployed a CQL-based tutor with statistically significant gains in natural learning. Recent: policy distillation with meta-RL, Robust DQN for stock market trading, and GNNs for prosthetic-limb gait. Seeking Summer 2026 ML/AI internship.
Work Experience
Graduate Researcher (PhD), North Carolina State University
Aug 2022 – Present
- Offline RL for ITS: Combined CQL with policy distillation from reference "good" policies; improved outcome prediction AUC by +0.06 using DR/WDR OPE on a 10-year dataset (12 problems; problem→step hierarchy).
- Meta-RL Transfer (PEARL): Adapted underperforming tutor policies using limited new interactions from multiple prior tutors; reduced cold-start regret in offline sims.
- Hierarchical RL & Delayed Rewards: Trained and deployed a CQL-based high-level tutor; observed statistically significant improvement in natural learning gain.
- Student Modeling: Built transformer/time-series student models to predict post-test scores and learning gain from trajectory logs.
Graduate Teaching Assistant, North Carolina State University
Aug 2022 – Dec 2024
- CSC 216/217 (Software Development Fundamentals): Fall 2022, Spring 2023
- CSC 522 (Automated Learning and Data Analysis): Fall 2023, Spring 2024, Fall 2024, Fall 2025
Software Engineer, Samsung R&D Institute Bangladesh
Dec 2018 – Aug 2022
- Developed JavaScript & Web API modules for the Galaxy Watch web app framework (Angular/React/HTML/CSS).
- Contributed to Samsung Internet Browser (Android) within the Android Framework stack.
Education
- Ph.D. (Computer Science), North Carolina State University, 2022 - Present, Advisor: Prof. Min Chi (GPA: 3.92/4.0)
- B.Sc. (Computer Science & Engineering), Bangladesh University of Engineering and Technology, 2013 - 2018 (GPA: 3.63/4.0)
Technical Skills
- Languages: Python, C/C++, Java, SQL, Bash
- ML/DL: PyTorch, JAX/Flax, TensorFlow, scikit-learn, Hugging Face, LightGBM
- RL: DQN/QR-DQN, DDQN, DDPG, TD3/TD4, SAC, PPO/TRPO, CQL, Decision/Trajectory Transformer, OPE (DR/WDR)
- Data/Systems: Docker, Kubernetes, ROS 2
- Other: Git, Linux, NumPy/Pandas, OpenCV, LaTeX