Technical Interviewer
San Francisco startup backed by Afore Capital, with $250K in pre-seed funding. I designed and run the technical interview loop, assessing the engineering candidates the founders shortlist.
Singapore · AI Engineer
I build AI systems that ship to production - computer vision pipelines, speech recognition, voice agents. Most of my work sits at the intersection of "make the model work" and "make it work reliably".
San Francisco startup backed by Afore Capital, with $250K in pre-seed funding. I designed and run the technical interview loop, assessing the engineering candidates the founders shortlist.
Co-founded non-profit with YellowRibbon.gov.sg, HTX, and Singapore Prison Service. AI-powered resume services now delivered to 600+ users. Collaborating with SPD (Society for the Physically Disabled); hosted a sharing session training career facilitators on using the platform. Awarded SGD 24,000 in funding by the National Youth Council. Signed 12-month MOU with YellowRibbon. Featured in The Business Times and Lianhe Zaobao.
Republic of Singapore Air Force (RAiD)
Built the full-stack system that analyses Air Traffic Control situations: an NVIDIA NeMo speech recognition and speaker diarisation layer (finetuned parakeet-tdt, canary-1b and NGPU-LM) feeding a Qwen3 reasoning layer served with vLLM, deployed to edge devices for real-time transcription in the control room. Built an LLM-powered internal search tool at a RAiD hackathon hosted at Microsoft's office.
Bifrost AI
Series A startup building synthetic data solutions for AI companies, backed by Sequoia Capital. Prototyped a zero-shot method with JinaCLIP to visualise missing data for NASA, BigBear AI and DSTA. Trained YOLOv6-lite for military tank detection with ST Engineering's Unmanned Vehicles Team; tuning for incremental learning lifted accuracy 16%+. Fine-tuned a StabilityAI VAE for synthetic-to-real domain adaptation.
GovTech Singapore
Video Analytics team, Data Science and AI Division. Red-teamed GovTech's AI Image Detector with custom-trained LoRAs. Lead AI engineer on a computer vision pipeline for MSO's OneService App on AWS (EKS, SageMaker), classifying thousands of daily municipal reports via zero-shot ensembles. Built real-time animal welfare detection for NParks with OWLv2, tuned by Bayesian optimisation (Hyperopt). Prototyped a novel object-detection explainability method using SegGPT and SAM.
State-of-the-art 5.99% Word Error Rate on the Air Traffic Control Automatic Speech Recognition (ATC-ASR) benchmark, with a parakeet-tdt-0.6b-v3 finetuned in under an hour on a single H200. Peaked at 200+ downloads a month, now averaging around 70
Co-author on Crowdsource, Crawl, or Generate? Creating SEA-VL, ACL 2025 Main Conference — annotated Myanmar-specific imagery and contributed to the writing
150+ member Discord / 3 workshops / 3 competitions / team ranked top 6% globally at Reply CodeTeen
Nationwide hackathon across all 5 polytechnics, sponsored by Google and AI Singapore
Bachelors in Computer Science, Minor in Mathematics
NUS Merit Scholarship — awarded to high-calibre freshmen on academic results, leadership qualities and co-curricular record
Diploma in Applied AI and Analytics
GPA 4.0 / 4.0
Lee Kuan Yew Award for Mathematics & Science (top tech graduate) / IMDA Gold Medal (top all-around achiever) / SP Scholarship