Motoki Wu

Work History

Salient (Applied AI Engineer)

I mainly work on self-improving models for flagging compliance issues for human and voice AI agent calls.

Cresta (ML Engineer / Data Scientist)

I worked on continued pre-training, post-training, and evals of LLMs for chat, search, and voice agents. Small models required extensive continued pre-training on domain-specific data (autocomplete, embeddings), while larger models needed post-training for response and query generation. For agents, the focus was on building strong evals and systematic prompt optimization.

Chegg (Lead Data Scientist)

I integrated WriteLab’s tech into Chegg’s writing product.

WriteLab (NLP Engineer)

I built the grammar and style detection ML engine for WriteLab. I independently came up with labeling functions that combine domain experts with machine learning.

FoxType (Co-Founder)

I built models to detect L2 language learner writing errors, including independently developing bidirectional language models that significantly improved performance.