I follow Symptom Driven Learning, use a four layered questioning process, and stay a generalist through speed learning to build good taste in problem selection before specializing.

A Very Opinionated Take on Year 1

Q: What did you do to get internships? (research/startup)

I did research work at the Language Technology Research Centre of IIIT Hyderabad from my second semester, followed a philosophy derived from expectancy violation and signaling theory, worked at a couple of small startups first to learn teamwork, and kept networking with strangers as the sole constant.

Q: What was your timeline?

Get bored, build something to prove a point, post it, and let whoever notices pull you into the next thing.

Q: How do I start X, where X = anything?

Read Scott Young's articles on Ultralearning, get out of analysis paralysis, hunt for resources recommended by people at the top of their field, and reverse engineer what they did before they got there.

July 16, 2026

A good read for anyone just starting university. Year 1 is different, there's no pressure to perform, no fear of failing. Instead, it's filled with room for unbounded exploration and rekindling whatever childlike curiosity you might have lost. It doesn't stay this way in the years to come.

Based on 600+ DMs on Reddit, Discord and Twitter I got in the last 2 months, I structured this essay as a Q&A.

1. How do you learn? #
I follow something called Symptom Driven Learning, and use a four layered questioning process. It fails for cases with no epistemic clarity. When I first started at university, I used a very crude plan to study, but I obsessively executed it for 6 months. My work is that of a generalist, hence I do speed learning, but I believe it will help me acquire a good taste in problem selection before I decide to specialize in anything.

I (unknowingly) built 25% of all the projects from here, much thanks to all the /g/entooman on 4chan. Here's what I did in the second half of 2025.

As for AI induced anxiety, I can only speculate. I feel slightly scared nevertheless. Most things are difficult, but not impossible.

Hence don't overthink, just start, and dive a little deeper. If you suffer from analysis paralysis, play around a bit first.

2. What did you do to get internships? (research/startup) #
I found it very enriching to do research work at the Language Technology Research Centre of IIIT H during my second semester (still ongoing as of July 2026), and I have come to believe that knowing how to do good research is beneficial in every step of life, from buying potatoes in a shop, to figuring out approximations for Determinantal Point Process to finding vulnerabilities in RP2350.

I followed a very simple philosophy derived from expectancy violation and signaling theories coupled with ambient social proof. I name drop a lot of jargon from unrelated fields, but that's what reading 2-3 volumes of Encyclopedia Britannica during your pre-teen years does to your brain.

If you're reading this before the next fall or winter, read this thread, start early to reap the most benefits during this time.

I worked at a couple of small startups at first to understand teamwork and how to follow deadlines. Now I work at a better place with better pay, but the work got increasingly research edged. In between, I had a rush of offers from a few SF based mid sized startups working on performance optimization, ML inference and (surprisingly) one aerospace company. There are two ways to interpret it: (a) get into somewhere small for exposure, and then pivot to bigger places once you have something to say for your "prior experience(s)", or (b) work on very hard and niche areas, and start better. Up to your discretion, MAANG (N for Nvidia) doesn't want freshmen who are reading this article anyway, but hey, those are for years to come. Among all these shenanigans, keep networking with strangers as the sole constant, at least that's what I did. There's a slim chance you'd be noticed by talent scouts, and be invited to attend networking events dressed as a house party, make sure to make great use of such cases.

PSA If you believe you're working on something ambitious, but circumstances (or location) make it difficult to connect with the right people, you can reach out to me, I'll be happy to get you in touch with the folks from Lagrange Point. Remember to pay it forward.

3. What was your timeline? #
Read this raw .txt file. In short, get bored, build something to prove a point, post it, let whoever notices pull you into the next thing.

4. How do I start X? where X = anything #
I'll encourage you to read Scott Young's articles on Ultralearning, and try to get out of analysis paralysis. Rule of thumb is to hunt for resources recommended or published by people at the top of their field, and to reverse engineer what a person of interest was doing or reading before they got into somewhere big. Google is your best friend. Try using the 4 layered questioning process with ChatGPT if you're still indecisive.

If you are curious, I collect all the niche resources from HackerNews, Lobsters, Reddit, Twitter and 4chan. Me and my friends took the liberty to make some of them public. Other than that, Vivek Galatage, Ben Dicken, Laurie Kirk etc. share great resources on Twitter. Having a good Twitter feed does help ;D

Ending notes
Apologies if I sent you the link to this essay without handing you the answers linearly. But burnout is a stupid bitch, and reading long form content is better in every way than getting a quick response. I still reply to well written emails, hit me up if you have something thoughtful to say.