Post PhD Reflections

I feel incredibly lucky to have had the PhD experience that I did. My advisors gave me a great deal of freedom to pursue my own interests, while providing invaluable guidance along the way. No PhD goes exactly as planned, and mine was no exception! At this point, I feel worthy of being able to provide at least some useful advice, and hope that I can help the next generation of astrophysics PhD students find their way as best they can.

Advice for Prospective and Early PhD Students

If deciding whether to pursue a PhD, I would give advice similar to what is (probably) given to aspiring professinal musicians or atheletes — pursue a PhD if you 1. really, really want to do it and believe it to be the best professional path for you in the next five to six years, and 2. have a sense that you’ll do well and be happy in that environment for the next five to six years.

Next, when looking for a research group, don’t look just at the groups whose scientific focus matches your current one — chances are yours will change! Look for an advisor whose approach you admire and a group whose environment you can see yourself being productive and happy in. There’s a lot of talk about “hands-on” and “hands-off” advising styles, but just pursue whatever feels most natural during the visit, rotation, or trial period.

Finally, once settled into a group, make the most of it! Get to know everyone within your group and your department. In my experience, there will always be people outside of your main group who you will learn a great deal from, and who may one day help you out when looking for advice, a new job, a letter of recommendation, or just a pleasant chat. Additionally, if there are additional skills you’d like to work on (e.g. machine learning, software engineering, teaching, instrumentation, etc.) make sure to pursue them in a meaningful way early on. If there are internships or fellowships that you find interesting, apply!

My Current Work at Scale AI

I’ve joined the Human Frontier Collective as a summer intern at Scale AI, learning the ropes of reinforcement learning with human feedback (RLHF). While I can’t share most of the details behind my work, I can say that I’ve had a great time helping advance the next generation of frontier LLMs, while leveraging my expertise in physics and astrophysics. It has been amazing to see the reasoning capabilities of LLMs advance over the last few years, and I believe that the work I’m doing this summer will play a part in continuing this trend.

— Tony