Lead ML Engineer @ Crunchyroll • Recommendations • AI Platforms • NLP Research
I'm a Lead Machine Learning Engineer at Crunchyroll (Sony Pictures Entertainment), where I build recommendation and personalization systems—and a centralized AI platform—for more than 15 million anime fans. My work spans architectures for training and serving language and multimodal models across production use cases.
Previously, I was an Applied Researcher in Natural Language Processing at USC Information Sciences Institute, where I worked on conversational AI for moderating harmful online discourse. That work led to a NAACL 2024 paper on language-model moderators.
I write about post-training AI models, LLM alignment, Recommendation systems, and the hidden quirks of systems we build and live in.
| Latest writing |
|---|
| Beyond the Numbers What happens when systems optimize the wrong things? |
| Policy Optimization Algorithms for Alignment A practical tour of PPO, GRPO, and DPO. |
| Gaming the System: Understanding Reward Hacking How clever shortcuts can derail language-model training. |
🌐 Read more on my blog →
Essays on Substack →
Have any thoughts about recommendation systems, ML platforms, or responsible AI? I'd be glad to compare notes. Interesting opportunities are welcome too.
