Work experience

The work I enjoy most starts before any code. I like talking to people as much as I like writing code: working out what's really slowing them down, and translating their pain points into a set of requirements.

I like building full-stack systems that are carefully designed, with AI as one part of them. The model is usually a small piece. Most of the work is the design around it: the data, the interfaces, and what happens when the model gets it wrong.

  1. Jul 2025 – Present
    1 yr 4 mos
    Singapore
    Read

    AI Engineer · Carro

    Leading development of Carro's customer-facing ticketing agents. Currently working on an internal multi-agent orchestration platform for department workflows.

    • Led development of Carro’s ticketing agents for 60,000+ monthly tickets across 3 countries. Worked with operations teams to turn their processes into agentic workflows spanning the customer journey (triage, inventory creation, appointments).
    • Developed an internal multi-agent orchestration platform that mirrors Carro’s departments, automating their workflows such as data retrieval and reporting, governed by role-based access controls for multiple users and roles.
    • Reduced LLM costs by 80% through harness engineering and intelligent model routing.
    • Designed an LLM-as-judge evaluation pipeline that validates agent changes before release, reducing SOP deviations by 30%.
    • Automated marketing outreach (customer engagement, follow-up) on n8n, driving 30% revenue growth for SG workshops.
  2. Jun 2024 – Apr 2025
    11 mos
    Singapore
    Read

    AI Engineer · AI Singapore

    Trained and deployed diffusion models that turn text prompts into 3D character motion.

    • Delivered a GenAI MVP for a gaming industry project, generating 3D character animations from text.
    • Built end-to-end ML pipelines in PyTorch to train and deploy diffusion models for text-to-motion generation.
    • Auto-labeled training data with multimodal video LLMs, improving text-to-motion metrics by 20% over baseline.
    • Optimized GitLab CI/CD pipelines, reducing testing and deployment time by 80%.
  3. Oct 2023 – Mar 2024
    6 mos
    Singapore
    Read

    Research Assistant · National Institute of Education

    Worked with professors to measure student participation in classroom recordings, using speech processing and speaker diarization.

    • Partnered with professors to build a speaker diarization system that measures student participation and learning outcomes.
    • Fine-tuned NVIDIA NeMo ASR models on self-annotated transcripts, beating open-source baselines by 30%.
  4. May 2023 – Aug 2023
    4 mos
    Singapore
    Read

    Machine Learning Engineer Intern · Department of Statistics

    Prototyped early LLM tools for government departments.

    • Prototyped NLP tools for several departments (classification, Q&A, report generation) using LangChain and GPT-3.5.
  5. Jul 2022 – Feb 2023
    8 mos
    Singapore
    Read

    Database Administrator Intern · Tessaract.io

    Worked on database maintenance and data migrations for law firms, at a legal-tech startup.

    • Maintained production PostgreSQL databases with PL/pgSQL and Python, migrating 300,000+ records for 10+ law firms.
  6. Feb 2020 – Aug 2020
    7 mos
    Singapore

    Professional Patty Flipper · Shake Shack

    • Ensured patties achieved optimal flip timing and acceptable levels of char.
    • Kept orders moving during peak hours.