03/RESEARCH
Research
Four research positions across scientific machine learning, LLM safety alignment, and additive manufacturing.
Interests
- Scientific machine learning
- Numerical solvers and learned models in the same system, where the analysis tradition sets the standard for what counts as correct.
- LLM safety and evaluation
- Evaluation design for failures that are fluent rather than loud, grouped by prompt family instead of collapsed into a single refusal rate.
- Low-latency systems
- Treating tail latency as a distribution to be measured rather than an average to be reported.
Appointments
Jul – Aug 2026
Nanyang Technological University
Undergraduate Research Assistant
Recalled to SC3DP as an Undergraduate Research Assistant with Prof. Paulo Bartolo.
Aug 2025 – May 2026
NTU CCDS
Research Intern
LLM safety alignment research with Prof. Anupam Chattopadhyay at the College of Computing and Data Science.
Jul 2025 – Mar 2026
MIT Julia Lab
Research Intern
Scientific machine learning research under PI Dr. Chris Rackauckas.
Jan – Jul 2025
NTU SC3DP
Research Intern
Singapore Centre for 3D Printing, under Prof. Paulo Bartolo.
Publications
A paper on LLM safety alignment was produced with Prof. Anupam Chattopadhyay at NTU CCDS between August 2025 and May 2026. Title, venue, and a link will be listed here once they are confirmed.
Citation pending confirmation
Research-adjacent projects
- Empirical Study of RoA Penalty Functions in NeuralLyapunov.jl
A controlled study of how region-of-attraction penalties shape stability certificates learned by physics-informed networks.
- Federated Whisper Aggregation Pipeline
Fine-tuning a production speech model across decentralised datasets without moving any data.
- MoE Compression of LLM Weights
Post-training compression of Mixture-of-Experts checkpoints, benchmarked for semantic drift rather than size alone.