RESEARCH AGENT · 2026
AI4Science Research Agent
Automates long research workflows such as literature review and experiment reproduction, supported by deterministic pipelines, durable state, and offline evaluation.
AI AGENT SYSTEMS · FULL STACK · SINGAPORE
LLM Agent Systems Engineer / Applied AI Engineer
Six years of full-stack engineering experience and currently pursuing an MSc in Applied Artificial Intelligence at NTU Singapore. I turn model capability into products: reliable agents, enterprise delivery, and deployable end-to-end systems.
Open to AI Agent, AI infrastructure, and full-stack AI roles
SELECTED WORK
Selected work spans reliable agents, research productivity, and multimodal AI, with verifiable outcomes beyond demos.
RESEARCH AGENT · 2026
Automates long research workflows such as literature review and experiment reproduction, supported by deterministic pipelines, durable state, and offline evaluation.
A job-search productivity tool combining browser automation and AI workflows, built around fault tolerance, recovery, and reliable execution.
A game-inspired productivity product that turns focus habits into dungeon progress, with synced tasks, progress, and achievements.
An emotional voice-cloning pipeline based on CosyVoice 3, using four-dimensional evaluation to identify semantic leakage and long-form failure boundaries.
Uses structured observation, evidence, and conclusion traces to turn image-emotion classification into explainable reasoning.
A self-collected, curated, and annotated image dataset for pet-emotion classification and interpretable computer-vision labels.
A game playground for curious minds. Drag people, scenes, and anomalies into a turtle-soup mystery, then compare how different models reason through it.
EXPERIENCE
iFLYTEK
Led end-to-end development, delivery, and commercial rollout for an online Office product serving 20+ enterprise clients. Built an enterprise-knowledge RAG system across retrieval, prompting, evaluation, and deployment.
Guangzhou Zhuowang Digital · China Mobile subsidiary
Led delivery of high-concurrency digital-marketing systems, owning service architecture, frontend framework analysis, engineering standards, and team collaboration.
CAPABILITIES
Tool use, MCP, ReAct, memory, workflow orchestration, LangGraph, Google ADK
Embedding, reranking, hybrid search, offline evaluation, and bad-case analysis
Python, TypeScript, Node.js, web applications, databases, Linux, Docker
PyTorch, Transformers, SFT, LoRA, multimodal inference, prompt engineering
EDUCATION
2025—2026
LET'S BUILD
Based in Singapore, interested in AI Agent systems, AI infrastructure, applied LLMs, and full-stack AI.