Reflections on 2025
If I had a dollar for every time I started a personal blog, I’d have four dollars by now. This post is the start to earn the fifth.
Table of Contents
- Motivation
- 2025
- 2026
- Now, let's look at Claude's advice
- General Observations and Learnings
- Closing Thoughts
Motivation
Reflecting back on the year that was, the "what went well, what didn’t, and what can be improved" kind of stuff, it struck me that besides The Nibble, I haven't really published anything long-form of my own, despite making fervent promises to myself and others that I'd do so.
The guilt of not publishing a blog and postponing it for so many weeks months was piling up and killing me inside, and I had to find my redemption. Enter Nate Soares' Replacing Guilt series of blog posts1. In just the initial few posts, something clicked: instead of "steering myself with guilt", I can transmute the guilt into resolve and "steer with resolve" instead, which is an easier emotion to process and handle. Hence, I am writing this post to take the observations from the last twelve months and set resolutions for the next twelve, and putting it out there to make myself more accountable.
2025
So, here’s everything I achieved in 2025 (in no particular order):
- Got a new job as an ML Engineer
- Shifted to Bangalore and explored Andhra cuisine2
- Reached 100 posts in Nibble + did a small Nibble meetup (100 weeks of continuous publishing!!!)
- Ran the first timed 10k of my life (race time: 01:10:50)
- Crossed 1k followers on Twitter
- Met some very interesting people, primarily through Twitter3
- Explored a lot of new places in India
- Did 40 push-ups in one go (discovered this limit in a random challenge)
- Got over a terrible lower back pain and (re)started working out
- Got myself a bicycle (check 2026 goals lol)
- Got rokafied
- Did a couple of small OSS contributions (primarily because of work, but now I see a lot of potential in continuing contributions)
- Read a lot - 10+ books, some manga, many research papers, and many, many blog posts
- Watched more movies, anime, and TV series
2026
And here’s everything I’d want to achieve in 2026 (again, in no particular order):
- More significant OSS Contributions
- Go deeper into RL and GPU programming
- Write more technical blogs
- Run a full marathon
- Cycle 50km straight
- Achieve some strength goals (compared to bodyweight - 1.5x Deadlift, 1.25x Squat, 1x Bench Press, 2.5x 45° Leg Press)
- Learn to play the harmonica
- Pursue at least one major idea to a complete project
- Read more books (12 this year)
And the following, which I have redacted for privacy reasons (will reveal the actual text at the end of 2026), so sharing the SHA256 commitment4 for now:
27add221ad5b5bcd06376eb7162d9005fe6eb54a2a47247edeb6904f799bdf8f712b5d8a4015d93e1f7a26e62e5110f7ad0f745fb4f0818319ae0ab130efeb214ba3d375f97579293b613c71e178ab27b5dcb4dd03231199ca7b11b456c9f289ff3af732c2b03bbfc20b086c8bc249d5df61b686d2501f057ed530fc207dc172
Now, let's look at Claude's advice
This year, I also chatted a lot with Claude. Hence, it was only natural to understand the patterns in my chats, which I can use to plan my 2026 better, and what it said gives me more reason to double down on writing.
The diagnosis hit uncomfortably close: "The gap isn't skill - it's legibility."
This is why half my 2026 goals are about output - technical blogs, OSS contributions, shipping a complete project. Not because I need to learn more, but because I need to make what I've learned visible.
General Observations and Learnings
This year, I thought a lot, talked with many people, read stuff, and applied many of the principles I learned in the process to my life. Here is a dump of some of them (non-exhaustive) that I have been thinking about. Some are my original thoughts, some are borrowed, and others are remixed, but everything is something I'm taking with me into 2026:
- Build a "good enough" heuristic for low-stakes choices. Don't get stuck in analysis paralysis for every little thing.
- Your responsibility does not end when you ship. (from here)
- The best time to do it is now.
- Claude Code with Opus 4.5 is a productivity multiplier. Claude App / Claude Code on mobile is the next best thing when you're on the go. Use it well.
- You can just do things.
- Working on things does not have to be binary (do/don't) - if you're limited on time/resources right now, brainstorm, collect ideas, and plan before actually starting.
- Say no more often and to more stuff that doesn't align with your goals.
- Say yes more often to things that do.
- Clear cognitive queues and be fast in prioritizing/dropping things that take your mental bandwidth.
- Don't postpone hard thinking or taking hard decisions just because they are hard. Know the opportunity cost of inaction.
- Don't have to follow X -> Y -> Z just because it’s the convention. Figure out the most efficient way to go from X -> Z, even if it means skipping Y and breaking the convention.
- Challenge yourself till impossibility.
- Sleep more, be kind to your body, and give it some rest.
- Never let your best times in life be behind you; don’t be smart in the past and a fool in the present.
- Hanlon's razor: Never attribute to malice that which is adequately explained by stupidity.
- Occupy mind with ideas, not people.
- Know the point of diminishing return and know when to go beyond it and when to stop before.
- Between two options - one bad and the other worse, choose the bad one without pain and then do it. Don't dwell on not having good options.
- Law of equal and opposite advice - For every piece of advice useful to one person, there is some other person who needs exactly the opposite advice. The internet is full of both.
- Don't burn your resources. It's a marathon out there.
- Desperation to goals with reckless abandonment and cautious deliberation, i.e., commit fully to your goals without hesitation, but also don't hesitate to change your entire life at the drop of a pen as soon as a better path appears.
- Have a goal so important that the social concerns drop away except exactly those that move towards achieving your goals. Believe it on a gut level. Think more about the task than what everybody thinks about you.
- Move faster without stagnancy. Act. Do not worry about breaking something. Try not to break something vital, but if you do, fix it and move forward.
- Everything is on the line.
- Know that it's okay to fail. Better fail and fail fast than have regrets.
- Love more.
Closing Thoughts
That's five dollars now.
This post isn't really about the blog count, nor is it about the guilt or escaping it. It's about resolve. 2025 was the year of internal compounding - learning, transitions, and whiplash. 2026 is when the leverage converts to output. Don't let perfect be the enemy of published.
See you at the reveal.
I listened to its podcast version because I find it more convenient.↩
Ironic that Andhra Cuisine is so popular in Bangalore despite Bangalore being the capital of Karnataka.↩
I have met 30+ people from Twitter in the last 2 years of being active. Pretty damn amazing to think about it.↩
commitment = SHA256(salt || plaintext)↩