I'm Alan, a current sophmore studying CS under UT Austin's Turing Scholars honors program and a researcher at the Learning Agents Research Group, where I work on reinforcement learning for vision-language-action models: teaching mobile robots to act on what they see and read.
I started programming at nine, making Mudkip-themed games on Scratch as @ac91305 (the avatar in the footer is the original).
Before research, there was Go. I competed as one of the top five youth players in the US, placed 3rd at AGA YoungLions and the US Go Congress in 2019–20, and represented the USA at tournaments in Shanghai and Osaka, training in the 5–6 dan classes while I was in China. What kept me at the board was never just the strategy. You can read a person through how they play, aggressive or patient or instinctual, before you share a single word of the same language. I was a kid memorizing joseki when AlphaGo and Lee Sedol traded Move 37 and Move 78; the animation on the homepage replays Lee's. I still run up games on KGS, and I'm thinking about starting a Go club at UT. Also, I managed to find a old game of mine displayed below (I swear I play better at in-person tournaments).
Research pulled me in the summer after ninth grade, at a UT San Antonio REU working on autonomous-vehicle simulation. That became my first paper. The next fall I joined the Intelligent Control Lab at CMU's Robotics Institute, and for two years I spent evenings on robotic assembly: three papers, one of them first-authored, and a lot of hours getting algorithms that worked in simulation to also work on a real FANUC arm. That work carried me to an ISEF Grand Award and Regeneron STS Scholar recognition, and the lab's Lego assembly testbed was part of what First Lady Jill Biden came to see at Mill 19.
Now I'm in Austin: fine-tuning robot foundation models at LARG, and fresh off a summer at Arm as an LLM research engineering intern on the Central Engineering AI team, where I independently developed a recursive language model approach and analyzed complex workloads to optimize uArch. The thing I cared about at nine is still the thing I care about: building stuff that has to work outside a vacuum, where reality evaluates your idea rollouts quickly.
When async off the clock: intramural basketball, lifting, teaching Sunday Chinese school (130+ students over four years), and a top-200-in-the-US Brawl Stars rank I'm still proud of.
