At Synapxe, I developed the HealthHub AI RAG chatbot for multilingual health inquiries, with a FastAPI backend deployed as Docker containerized apps. I also designed and deployed an agentic chatbot for vaccination eligibility checking and appointment booking, showcased at the AI Accelerate 2025 Conference.
At Shopee, I iterated on research-paper neural network architectures in PyTorch for chatbot intent recommendation. The deployed models improved customer satisfaction and CTR by up to 1.7% in A/B tests, and I developed large-scale user data processing pipelines with Apache Spark.
At DSO National Laboratories, I designed an agentic workflow for LLM-guided evolution of agent harnesses and ran ablation studies across public datasets. The research was accepted by the NeurIPS Meta-Agents workshop.
As a Volunteer Developer for Better.sg (Schemes.sg), I contribute to an Agentic RAG chatbot that matches people with schemes and serves over 400 active users. I’m also developing Firebase functions that automate link and content checks, saving volunteers up to 60% of scheme database maintenance time.
