I'm building Smithery to orchestrate the emerging agentic internet.
During my teens, I taught myself programming and launched Minecraft mods that reached over 5 million downloads.
In 2018, I joined UC San Diego's PhD program to research neural networks for music and natural language generation. 2 years later, I dropped out to co-found Jenni, an AI academic writing copilot. We scaled to 300K+ monthly active users and $7M+ annual recurring revenue before my exit in 2024.
I'm curious about the nature of intelligence. I believe we won't truly understand ourselves until we can build machines that think like us.
I angel invest in pre-seed and seed-stage startups. I'm most helpful in areas involving AI, technical architecture, product-led growth, and product design.
Huanru Henry Mao (2022) Fine-Tuning Pre-trained Transformers into Decaying Fast Weights. 2022 Conference on Empirical Methods in Natural Language Processing.
Huanru Henry Mao (2020) A Survey on Self-supervised Pre-training for Sequential Transfer Learning in Neural Networks. Arxiv.
Huanru Henry Mao, Shuyang Li, Julian McAuley, Garrison Cottrell (2020) Speech Recognition and Multi-Speaker Diarization of Long Conversations. INTERSPEECH 2020.
Thomas Bachlechner*, Bodhisattwa Prasad Majumder*, Huanru Henry Mao*, Garrison W. Cottrell, Julian McAuley (2020) ReZero is All You Need: Fast Convergence at Large Depth. Proceedings of the Thirty-Seventh Conference on Uncertainty in Artificial Intelligence (UAI 2021).
* Equal contribution
Huanru Henry Mao, Bodhisattwa Prasad Majumder, Julian McAuley, Garrison W. Cottrell (2019) Improving Neural Story Generation by Targeted Common Sense Grounding. 2019 Conference on Empirical Methods in Natural Language Processing.
Chris Donahue, Henry Mao, Ethan Yiting, Garrison W Cottrell, Julian McAuley (2019) LakhNES: Improving multi-instrumental music generation with cross-domain pre-training. 19th International Society for Music Information Retrieval Conference.
Chris Donahue, Henry Mao, Julian McAuley (2018) The NES Music Database: A symbolic music dataset with expressive performance attributes. Proceedings of the International Society for Music Information Retrieval Conference (ISMIR-19).
Huanru Henry Mao, Taylor Shin, and Garrison W. Cottrell (2018) DeepJ: Style-Specific Music Generation. 12th IEEE International Conference on Semantic Computing.