Scholar
Nikki Lijing Kuang
Google Scholar ID: XYhmg74AAAAJ
University of California San Diego
Citations & Impact
Citations
92
H-index
7
i10-index
4
Publications
16
Co-authors
12
list available
Publications
Resume (English only)
Academic Achievements
- Published multiple papers, including 'Towards Personalized Language Models via Inference-time Human Preference Optimization' (NeurIPS 2024 AFM), 'Diff-BBO: Diffusion-Based Inverse Modeling for Black-Box Optimization' (NeurIPS 2024 BDU), 'Log-concave Sampling from a Convex Body with a Barrier: a Robust and Unified Dikin Walk' (NeurIPS 2024), and more. Invited to give talks at various academic conferences.
Research Experience
- Interned at IBM Research, Amazon, and Honda Research Institute, working on LLM for personalization, RL for ranking and recommendation systems, and robotics. Experienced in fine-tuning LLMs and reward models, designing CoT prompting and reasoning frameworks, LLM decoding, and training R1-style reasoning LLMs using RL (e.g., PPO, GRPO).
Education
- PhD candidate in Computer Science at UC San Diego (expected 2025), advised by Prof. Yian Ma; MSc in Computer Science from UC San Diego (2020).
Background
- Primary research interests span reinforcement learning (RL), foundation models, and Bayesian inference, with a focus on addressing fundamental challenges in sequential decision making under uncertainty. Recently, particularly interested in LLM alignment and reasoning, exploring how RL plays a role in these topics. The goal is to design provably efficient and practical algorithms with performance guarantees, achieving both statistical and computational benefits.
Miscellany
- Received awards such as the NSF AIVO Travel Grant. Served as a reviewer for several international conferences (e.g., NeurIPS, AISTATS, AAAI, ICML, ICLR, ISIT) and journals (e.g., IEEE Transactions on Circuits and Systems for Video Technology, IEEE Transactions on Information Theory).
