March 2021: Building

Optimism is building that the end of the pandemic is here in my circles. I am thankful to be in an area where vaccines have been available (compounded by working in the education system still).

My work has been in a big building phase. Building code, building ideas, building relationships.

I have a few more papers under review, a few more approaching submission, and a couple I am excited about gaining momentum in experiments. Reflecting on this progression is how long it takes for productive relationships to engage. For reference, my first and second years of grad school I was on 1 paper. In 2020 I was lucky to be on 5, and in 2021 it may reach 10. Take this with you to any project or company, but know that sometimes sticking out the full productive 3-6 years may not be the best choice.

Some of this is from overworking, but rewards come at times asymmetrically delayed from the input. Personally, I am building the right foundation for growth as a happy human. Graduate school is a place not conducive to this, but I am slowly getting my feet back under me (it was not great to enter the pandemic in a graduate school depression phase). I am more than pleased with my success as an academic, but I am openly still learning to be my most present, happy, and engaged self. With Twitter I have been continuing to meet new graduate students and professionals in my area. It took at least 6 months for the algorithms / impressions of others to engage with me, but now it is an asset. Not all jobs engage so strongly on Twitter, but academics love it because its a competition ground for their ideas in a plain text form.

I expect April to be similar, where some of my building code will be the launch of two open-source simulators (model-based RL and swarm control). From May, I will reset some of these processes with an internship at DeepMind.

Some done things:

Media & Academic

  • I was on the TalkRL podcast on Model-based RL, Trajectory-based models, Quadrotor control, Hyperparameter Optimization for MBRL, RL vs PID control, and more! This was quite a fun episode and a well done introduction to my research.
  • I gave a talk back at Cornell! It was a pleasure to return to my alma mater and their Robotics Seminar.
  • I added my Bookshelf to my website (with curated lists). I am excited for a visual and time-indexed way to track my books.
  • I helped UC Berkeley EECS implement a new professionally moderated peer support group, and it went super well. I am hyped about continuing to improve the global picture of mental-health for students.

Blogs

  • I am working on two posts that I want people to help read drafts: debugging model-based RL code and how to approach the graduate school application process.

Democratizing Automation

My Substack has been continuing to grow (finally in 200+ club). It is mostly covering reinforcement learning (RL) and robotics recently, with more on RL coming soon. This was a busy month of work and the blog has taken the back seat while I am working on peer-reviewed material in the same space.

See you next month! Search for more!