main account of @amuldotexe & @thatinrust
heap
Joined January 2023
- Tweets1.3K
- Following487
- Followers1.5K
- Likes6.4K
While am working and shipping to production on backend, is there some cohort based course which can help me learn via real hands-on projects - more deeply about deployment across azure, aws - accelerating my learning on the system design trade-offs side?
crio was one, any others? can I have that learning without a cohort based course
I represented my state in U14/U17 age groups table tennis tournaments & soon learnt that while competing in sports is seen as something noble, it barely makes money even for the most talented folks out there
The TLDR is sports doesn't make money. The best people can do is get a government job.
Ok try to see it this way: everybody I played with who didn't pursue proper engineering or similar paths is not even in a government job. They're likely just teaching a sport in some of these societies and you can say that maybe that is what they like doing for a living. I think they were as smart and driven as I was and I just got lucky that I got into tech or I got into fields which are more prosperous.
This is the point here: similar people, similar background, extremely different outcomes. You could say some of it is just a single variable called "what prosperous field did you pursue?"
I am completing 16 years in the field and let's say I did a graduation of 4 years so it's 20 years from being a teenager. As a teenager all of us were from similar backgrounds. All of us were similarly smart in IQ and hard work. I would argue some of them were more hardworking than I was, more disciplined, and some of them were genuinely very capable of a level of devotion which can put a lot of good craftsmen in our industry to shame.
They were so capable of controlling their impulse. They were so capable of putting in long hours. They were so capable of persisting for months over a single pursuit. It's unreal how competition shaped us. It helped us become better. It helped us believe that a goal is worth months or years of constant pursuit. That growth is real. That what seems magical today is just a lot of hours put into the craft.
Those are very beautiful traits to learn and they had those traits and they leveraged them and learned that pattern of deliberate practise of seeing the exponential curve of learning, reaching a point where it feels like magic. They saw it day in and day out. Yet when you look at business and money most of them didn't make it so well. I always wondered what if they had taken a different path?
In my case my parents, despite their other failures as a parent, definitely did push me in the right direction by not allowing me to continue playing sports and instead shut it down when I became a teenager and pushed me to study and prepare for entrance exams.
I hated them for it initially but life has shown me that it was an infinitely better decision than letting me continue in sports. I am very lucky that they did not risk my career that way.
Another interesting piece here is that a lot of players who even reached higher levels than mine, maybe became India's top 10, actually would have had injuries that pushed them back by decades. There was the risk of a single physical injury ending the game for you. In India apart from cricket almost all other games have very little audience and fame and money that comes with it. At least that was the case till 2000. I think now that is changing and that is good that we are diversifying into n number of sports.
The essence is that sports is a great entertaining event and that is where a lot of people confuse it with some noble pursuit. It is not a noble pursuit. It is a pursuit of a good show that viewers will pay for in some way: the thrill of it, the excitement of it, the joy of watching people compete.
Why it feels unfair is that most people who devote themselves to sports are generally very smart and, more importantly, very disciplined. You cannot achieve a certain level of outcome without a certain level of practise and discipline. That is kind of what makes you strong and fit for a lot of good corporate jobs but it also comes at a cost.
In terms of zero-sum games, the competitive nature of sports often leads to people harbouring a deep scarcity mindset because there are very few winners. Now that is helpful for some business positions but for some it just becomes a liability, especially if you are in a creative field like tech, where it's not about winning. It's about creating something cool or differentiated while believing that many cool things can simultaneously exist.
So I am arguing that if you find good sportsmen and can train them for a particular, say, craft, they can be good employees. The downside though is that a lot of people come to sports because they didn't want to study. They didn't want to put in the hard work but that again depends on the context. Some of them, if they would have been given the chance to study, would have done pretty well but they just weren't from such families or such places.
amul.code retweeted
@nikitabier this is such a good feature!
amul.code retweeted
I am working with @sheetalssr to convert the best of my 60k tweets into a book - these are the chapters list - sample - would love to know thoughts of folks who have read me for a while
The 34-Chapter Table of Contents
Prologue: The Apology
Part 1: The Country You're Actually Competing In
1. 25 Million
2. Geography Is Destiny (But Not Forever)
3. The Availability Bias Trap
Part 2: The Low Risk Career Map
4. The One-Variable Career Formula
5. Dev First
6. The PM Trap
7. Brand Names Are Future Money
8. Always Interviewing
Part 3: The Money Lessons You Learn in 30s
9. Salary Growth Beats Investment Returns (Until It Doesn't)
10. The FIRE Miscalculation
11. The Optionality Premium
Part 4: Health Is the Compound Function
12. The Six Months I'd Trade Everything For
13. NSDR and the Art of Slowing Time
14. Sleep, Not Hustle
Part 5: How Advice Actually Works
15. The Principal-Agent Problem in Your DMs
16. Likeable Advice Is Not Effective Advice
17. The Hypothesis Habit
18. Why I Won't Advise You If You Have Under Three Years of Experience
Part 6: Building Conviction Slowly
19. Marination
20. The 6-Year Joke
21. Build It, Don't Borrow It
Part 7: The Long Game (Operational)
22. 50 Mocks Before Your First Interview
23. Open Source as Proof of Work
24. The Trust Pipeline
25. Micro-Circles Over Mass Audiences
Part 8: Low Drama as a Philosophy
26. The Drama Tax
27. Effortless Over Hyper-Intentional
28. Low Intensity Language as Competitive Advantage
Part 9: Slowing Down Time
29. The Pre-7am Tea
30. The Work Station
31. "All We Can Do Is Slow Down Time"
Part 10 This Book Is Not For You It Is For Them
32. Competing Gracefully In 25MM Cohort
33. Don't Make Them Your Project
34. The Passport Question
amul.code retweeted
✨ New launch: more than 2000 product leaders, CxOs, founders have taken my Product Sense + Strategy course. After taking it, many have asked for a self-paced format for their teams
Request priority access for your team via this Google form (takes 2 min):
forms.gle/9v16SCrtRE64sTyB6
amul.code retweeted
Today first time I got a monthly payment for a developer contract, one of 480 instances, iykwim
tl;dr of my 45 days of Rust OSS sabbatical
- first b'day since 2011 when I am not gainfully employed
- money in bank feels good, but fresh salary feels much better
- how even if you have finances sorted, being without a logo feels odd to the 90s kid
- product to systems programming does not have precedence, so hiring is slower
- LLMs have increased the speed of acquiring systems programming knowledge
- Reading code is necessary for deeper quality contributions to large repositories
- larger + deeper codebase contributions are more important than individual libraries with viral potential, the individual personal projects are infact a distraction
- FDE exploration and interviews were a distraction, lost a few weeks on that direction, converged on the right narrative "senior product manager turned systems programmer"
- do fewer PR merges but make sure you understand your code, OSS maintainers are now
It's perhaps my first birthday in the last 16 years when I am not in full-time employment. Those of you who have always seen my posts on celebrating the 30th of the month salary credit thing would relate to this by default. Our careers are more non-linear than we like to admit, or at least for mediocre people they are not always an upward graph, and that is not adequately captured in the representation we see on the internet, I think. That is where I think I always had a gratitude for every month that you were remunerated so well for pressing keys, as I call it. All these years of planning everything very well financially, but the part that you are not prepared for is this feeling of not-being-of-value to the world. You have adequate finances, but you still feel like, because you are not paid a remuneration for this month, you are not of value to the world. And I do agree that now a sabbatical is generally not looked down upon as much and is often seen as a time to prepare yourself for a new gig, recharge yourself. Coming from the 90s, it surely feels very odd because you tend to define yourself by having a logo that you are working for.
I think in my particular case, while I have a reasonable amount of conviction, the precedence of a 10 years product manager pivoting to a pure systems programming role is not yet established. I would be surprised if there is somebody who has done this before without, say, relevant training. But I believe that systems programming is going to be a very fast-growing domain. Something which was not possible, say, even two or three years ago, because the LLMs are going to make code generation for systems programming significantly faster, even though the amount of context needed to write the code will still be non-trivial.
The part that excites me the most is the compression of time needed to acquire systems programming knowledge, because previously it would be difficult to marry it and understand concepts from books. A lot of that stuff is really dry, but with LLMs you can ask it to explain the same thing, the same code, again and again and again. That is why some of us like me are able to inch by inch build up our knowledge in systems programming.
More recently, though, my bigger challenge has been distraction. Almost every day I feel that there is some repo that I could have created which would have gone viral. I also see that those particular repositories are viral for a few days, get a lot of stars, but then after that nobody is working on them because the next cool thing has come in and people are using that. The shelf life of software products is decreasing, and only a few software products are actually surviving longer windows. That is where I also think anything on the front of systems programming will probably last longer if it is able to find a true 10x improvement in the existing system.
Today marks the 45th day of my Rust OSS Sabbatical
I think one of the major mistakes I made was going after the forward deployed engineer role or the FDE role. I should have had the conviction to directly go off to the systems programming roles
I think the surface area for Forward Deployed Engineer is very high, and also the depth is lower - I think the key reason I pursued FDE was because it seemed to align with my previous experience of product management. I think that's a classic mistake. You carry the baggage of whatever you have done previously, and it's only later that I realized that this is not the direction I should be taking, but yes, I lost a few weeks on that distraction
One of the challenges I am working on is learning to read a lot of code quickly enough so that I can make sense of it and review it, because code generation is reasonably solved in terms of ingesting the right context for what you want to write, but the verification is very subtle. It needs a lot of domain context, a lot of codebase context, and a lot of tribal knowledge of whatever you are contributing to. That is something which you cannot acquire in just a day.
So my current focus is on mastering code verification - and this means that for me higher ROI is in sticking to a particular language and going deep in that, hence Rust is my bet on that game that at least Rust codebases I will be able to build context on very quickly
The key uncertainty is in terms of precedence - everybody I speak to has probably never heard of somebody making such a pivot, and that obviously is not very encouraging. Doing it as the first person can be a nice dramatic narrative, but it also means that maybe the odds are stacked against you, or maybe the averages are not in your favor. But then again, I have to remind myself that my context is different. I've been planning for this work for quite some time. I have been preparing my knowledge base and my mindset of learning in this direction for quite some time, so maybe it is just a matter of a few hundred hours of effort before my product-market fit is discovered
You know, I could be painting this as a story that, yes, I must do it because it has not been done before, but that is an idiotic line. The reason I want to do it is because I genuinely think a system programming role suits me better and will give me a good quality of life, because I will enjoy it far more and work will feel less like work. Not to mention it will also ensure that, for somebody like me aging and going into the 40s, it will create a very consistent form of employment. The skills become more tangible than, say, a general product management role, which is very intangible, and the demand-supply gap is very different or unpredictable
Systems programming, though a niche, is still very, very measurable. In fact, open source software is a boon in that sense because it literally allows you to ship code that is viewable, that is vouchable, that is scrutinizable in a very tangible way.
I've been reaching out to employers, saying, "Hey, if you want an open source library that I can build for you, tell me a problem statement and let me see if I can build something out in Rust for you which might be of help to you."
The key idea is that an employer might benefit from the open source tool that I create, and I will obviously be able to claim that this is a tool or a library which is already a good product-market fit and is validated by somebody
But here is the thing about your personal projects: there is a very low probability of them going viral. Even if they go viral, the virality in today's time is very, very ephemeral. Hence, the main job for reputation building is now shifting over to large open source code bases which are already proven in terms of product-market fit, are probably under the Apache or a similar umbrella, or have been funded already. Anything else is probably a distraction right now. Even if you think your own personal project is going to go super viral and change the world, maybe in the interest of time the key idea is to contribute to a larger code base
The contributing to a larger codebase thing is also very relevant in terms of how an employer might see you. If you can grow a large codebase and make sense of it and start contributing effectively very quickly, that is also a proof that you will do the same at work, because that is what happens. You rarely build new stuff. You come to a large existing codebase, grow all the context deeply, and then, in sync with that, you kind of find those areas to do small fixes and improvements. You slowly build a threshold amount of knowledge that helps you solve bigger problems inside that codebase. Hence, reading the codebase, understanding the codebase, and grokking the codebase is almost everything.
--
All in all, it has been a meandering journey, but every few days I find more clarity in what to focus on next
for the next 14 days - it is "grokking the large codebase"
will you posted on how that goes ...
amul.code retweeted
When I interview for a company my objective is always to score the offer first and then judge the company - because my judgement might be an emotional upheaval but can lead to me "self-sabotaging" the interview process;
so even if I am a little annoyed with the process and the experience, I focus hard on making sure I put my best self in the conversation - we can always critique the experience later - but this is the only moment to convince people to pay top dollar for us;
observe don't judge
till you get the offer
amul.code retweeted
📈New milestone: crossed 5000+ students
in the past 3.5 years
over 3 different live courses I’ve taught.
I absolutely love teaching and I often get the question “but Shreyas, don’t you feel the itch to do something that has greater impact than teaching?”
It is a valid question (as in, I don’t blame whoever might be asking or wondering), but it comes from a place of not seeing what I see: thousands of highly talented leaders who, every day, are applying what they’ve learned to their products, their strategies, their proposals, their leadership approach, and their careers.
I don’t do this work primarily for the impact (something I learned only in my 40s: when I do it for love and from a place of mastery in what I do, the impact takes care of itself). But every day, I get notes from students about the staggeringly compounding impact on their companies, teams, and careers. Which means I get to smile every day 🙌
To every student who’s been part of this journey thus far: thank you for your trust, affection, and for sharing your experience 🙏
amul.code retweeted
🆕I am starting to do more Executive Coaching now that some more space has opened up in my personal life. Now coaching 3 founders (this is different from my startup advising work) and I am now opening it up for senior product leaders.
Details are here: maven.com/shreyas-doshi
Premature optimization is the root of all evil
- Donald Knuth
Hypothesis:
It's easy to attribute your success to yourself, because the mind doesn't want to know the 10 other variables which played their part
This Post is from a suspended account. Learn more
amul.code retweeted
This recent @bchesky podcast with @patrick_oshag belongs to the hall of fame of tech podcast episodes. Very likely the deepest founder conversation after the Steve Jobs lost interview in the 90s.
Currently it has 90k views; it deserves 100X more views.
youtube.com/watch?v=eURcW5_u…
