Define and implement the "Definition of Done" for AI using quantitative evaluation. Use evaluation tools (e.g. DeepEval) to score models on Faithfulness and Relevancy. Use these metrics to validate model upgrades and cost-optimisation experiments
Collaborate with data scientists to improve feature selection, feature engineering, and enable end-to-end AI workflows from model training to deployment and monitoring.
Engineer agent prompting and context management systems: system prompt architecture, dynamic context construction, prompt versioning, and evaluation-driven prompt optimisation using automated and human evaluation frameworks.
The TikTok Live Revenue team owns the core monetization business of TikTok LIVE, including gifting, fan clubs, interaction and growth products for creators and viewers worldwide. The team invests heavily in AI engineering, building its own agents and developer toolchain, which are already applied to feature development, issue triage, testing and production monitoring.
Vertical Search Team: Build a full-stack search engine serving billions of global users, covering vertical scenarios such as photo-text search, user search, live search and music search. The scope also encompasses AI-powered search for standalone apps including Drama and Lemon8. We leverage LLM, NLP, and multimodal understanding technologies to iteratively enhance search quality and system performance.
4+ years as a software engineer, data engineer, or AI/ML practitioner writing technical design documents, data pipeline documentation, or API specifications