Jul 2026 — Present
Singapore
AI Research Student
WING NLP Lab, National University of Singapore
- Researching test-time scaling for foundation models with Prof. Min-Yen Kan, with a focus on inference-time self-improvement and iterative refinement.
Samuel Then
Computer Science at NUS · AI research and software engineering
I work on efficient multimodal learning, computational imaging, and reliable software systems. Currently researching test-time scaling at WING NLP Lab.
Experience & Research
Jan 2026 — Jun 2026
Singapore
Computer Vision Algorithm Intern
Sony Semiconductor Solutions
- Developed video-reconstruction models for photon-starved SPAD sensors, reaching 37.1 dB PSNR: +6.5 dB over the retrained baseline with about 30× fewer parameters and 248× fewer FLOPs.
- Derived finite-sample Cramér–Rao lower-bound analysis for motion estimation and built a physics-accurate stochastic sensor simulator.
Sep 2025 — Dec 2025
Seattle, USA
AI Research Student
Paul G. Allen School of CSE, University of Washington
- Studied visual tokenizers for multimodal models with Prof. Simon Du; trained six VLM combinations on eight L40 GPUs and evaluated them across 15 benchmarks.
- Built a 9-backbone × 6-probe evaluation suite spanning CLIP, DINOv2, SigLIP2, MAE, and VQ tokenizers.
May 2025 — Aug 2025
Singapore
Software Engineer Intern
Central Provident Fund Board
- Architected an enterprise migration pipeline for 75,000+ legacy files, processing documents about 50× faster than the previous workflow.
- Built an agentic sentiment-analysis platform and presented strategic insights to senior leadership.
May 2024 — Aug 2024
Bandung, Indonesia
Software Engineer Intern
Cyber Army Indonesia
- Engineered a CVE scanner with 50+ detection rules and prototyped cyber-threat-intelligence monitoring tools.
Jan 2024 — Apr 2024
Singapore
Software Engineer Intern, AI
NUS Smart Systems Institute
- Co-built TOM, an open-source wearable AI platform presented at UbiComp 2024, integrating YOLOv8 and language models into smart glasses.
Apr 2023 — Jul 2023
Kuching, Malaysia
Backend Engineer Intern
Rajang Digital Solutions
- Deployed named-entity recognition in an information-extraction workflow, improving system accuracy by 30%.
- Engineered document-processing and translation automation for financial documents, improving client satisfaction by 50%.
Education
2023 — 2027
National University of Singapore
Bachelor of Computing (Honours) in Computer Science · Second Major in Mathematics
GPA 4.89/5.00, Highest Distinction · AI Focus Area Certificate of Distinction · ASEAN Merit Scholarship · Dean's List (3 semesters)
Fall 2025
University of Washington
Student exchange at the Paul G. Allen School of Computer Science & Engineering · Dean's List
Recognition
Singapore National AI Student Challenge
3rd Place
NUS STePS 26
2nd Place
AWS Build On ASEAN
2nd Place
NUS Dean’s List
Three semesters
NUS ASEAN Undergraduate Merit Scholarship
Merit scholarship
Singapore Mathematical Olympiad
Bronze Award
Singapore Physics Olympiad
Bronze Award
Singapore Chemistry Olympiad
Bronze Award
Teaching
AY26/27 Semester 1
Current
Teaching Assistant
NUS Computing
- CS2109S — Introduction to AI and Machine Learning
- PI3880 — Advanced Technology Attachment Programme
AY24/25 Semester 1
Previous
Teaching Assistant
NUS Computing
- CS1101S — Programming Methodology I
- Received a 4.9/5.0 student rating.
Leadership
Director of Software Development
NUS Fintech Society
AY25/26
Product Associate
Google Developer Group
AY25/26
Event Director
NUS Golf Club
Aug 2024 — May 2025
Interest Group Leader
Pioneer House Table Tennis
Jan 2024 — May 2025
Co-Founder
Sarawak Student Initiative
Apr 2022 — Dec 2023
Selected Projects
Steel Demand Forecasting Platform
AI Singapore National AI Student Challenge · 3rd place
Reduced forecasting error by 38.6% over baseline and built an LLM-assisted analytics dashboard.
Sarcasm Detection & Fact Verification
Natural language processing · NUS STePS 2nd place
Reached 0.938 F1 on sarcasm classification and 0.901 F1 on fact verification.
n8n Community Nodes
Open source · TypeScript
Published nodes for in-memory Excel operations and CrewAI agents using n8n tools and workflows.
NER Labelling Platform
Next.js · Supabase
Built a collaborative annotation platform with live sync, role-based access, and support for Sarawak dialect resources.
Technical Skills
- AI
- PyTorch, computer vision, NLP, LLMs, diffusion models
- Engineering
- Python, TypeScript, Java, C, PostgreSQL, Docker, Terraform, CI/CD
- Languages
- English and Chinese (native)