Curriculum Vitae
Research at the edge of reliable language models.
Machine Learning Engineer at Oumi PBC. I work on post-training (SFT, RL, distillation) and spend my research time making open models capable and reliable.
ICLR ’25 Oral · first-author
4 Publications
4.85/5.0 cGPA · highest distinction
Top 3/600 A*STAR Talent Search
Experience
Machine Learning Engineer
- Own end-to-end development of on-policy distillation for the enterprise platform, from training methodology to backend APIs, frontend workflows, and billing integration.
- Lead training science for enterprise fine-tuning: experiments and production-grade recipes for SFT, RL, and distillation.
Machine Learning Engineering Intern
- Led ConseqBench, benchmarking how AI agents respond to risks in real-world environments.
- Built RAG + GRPO and web-agent SFT demos for Project Inspire, used in company pitch materials.
Senior Research Assistant
- Developed Ground-GRPO, an RL-based fine-tuning method for grounded citation generation in LLMs.
- Designed reward signals that reduce unsupported citations and improve attribution quality in RAG-style generation.
Master’s Student (Research)
- First-author paper (Trust-Align) on LLM trustworthiness metrics and alignment, accepted at ICLR 2025 (Oral); 40+ GitHub stars.
Bioinformatics Research Intern
- Independently built a Python/R workflow to identify therapeutically significant gene sets; co-authored a manuscript in Nucleic Acids Research (IF 16.6).
Machine Learning Research Intern
- Cut brain-signal-decoding latency 45% with Kalman filters and RNNs on resource-limited BCI chips, without losing performance.
Bioengineering Research Intern
- Analyzed protein structures in multidrug-resistant bacteria; 3rd place at A*STAR Talent Search among 600 projects.
Education
M.Eng (Research), Computer Science
- Thesis on reliable RAG systems in NLP.
- Completed three graduate-level courses in first year to secure a $15,000 research grant.
B.Eng (Hons), Computer Science
- cGPA 4.85/5.0, Honors with Highest Distinction.
- SUTD Global Distinguished Scholarship (top 10%).
Projects
- MediRAG: biomedical RAG pipeline using BGE, PubMed embeddings, and FLAN-T5 query expansion for better QA accuracy.Sep 2023
- Maze AI: CNN, autoencoder, and RL models for maze solving on 120k mazes; best AI course project.Sep 2023
- Nugen Studio: advanced RAG educational chatbot; custom Figma/React front end; 30+ user interviews, 20 product tests.2022 – 2024
Awards & Service
- Cloudsine Artificial Intelligence Award: top students in 50.021 Artificial Intelligence.2024
- High Performance Computing Innovation Challenge: 2nd place overall.2022
- A*STAR Talent Search: 3rd place of 600, for exceptional research ability.2019
- Climate Science, Author: co-authored a greenhouse-effect course for a charity reaching 40+ countries.2020
- SUTD Rotaract Club, Professional Development Director: led a university-wide forum with CEOs, senior Rotary executives, and 50+ students.2020 – 2021
Skills