About me

[linkedin] [github] [email]

Hi there! 😎 I'm Keyu Tian (田柯宇), a second-year Master's candidate @ Peking University under the guidance of Professor Liwei Wang.

I'm deeply passionate about advancing AI algorithms and systems, towards the ultimate goal of artificial general intelligence (AGI). My focus is on 1) tackling challenging open problems in Deep Learning, and 2) enhancing DL algorithms and systems for optimal efficiency and effectiveness.

I strongly believe that self-supervised learning and reinforcement learning with search are the most promising ways (at least for now) to achieving AGI. I explore research in these two areas, with works published at NeurIPS, ICLR, and more.

I have a strong background in algorithms and systems, with extensive familiarity in C/C++/Python and their underlying mechanics. Also ACM-ICPC trained. I take particular pleasure in optimizing my codes for supreme efficiency when coding with C/C++.

My extensive experience in engineering and teamwork has honed my ability to write readable, maintainable, and extensible code using various design patterns and principles.

Check out my linkedin page for more information.

Fun Facts & Hobbies

My first encounter with AI was through the animated series Code Lyoko, which I've loved since childhood. The series culminates in humans defeating an evil AI using a multi-agent system -- quite interesting.

I'm passionate about astronomy and a long subscriber to "Astronomy Enthusiast (天文爱好者)" magazine. I read a lot of science fiction novels, especially those are related to the universe. "The Three-Body Problem (三体)" definitely an amazing one and I like it best.

I'm drawn to music with a strong beat, such as Michael Jackson's R&B, The Cranberries' Rock, and Aphex/Recondite/Echospace's Techno.

Songs I've listened to for at least 5,000 minutes:

I'm also a big fan of StarCraft. Have been playing SC I/II for at least 15 years.

And finally, my motto:

Talk is cheap. Show me the code.