Lester Kim

AI Engineer | Helping B2B teams ship AI products w/ confidence using evals | ex Head of Eng @balconytech | ex-SWE @MagicEden @krakenfx | AB Physics @Harvard

[Lester Kim @LesterCKim]

@growing_daniel Of course. The core skill is software engineering, and AI/ML engineers are subsets. Plenty of other subdisciplines like blockchain development, DevSecOps, networking, embedded, etc.

[Lester Kim @LesterCKim]

@GergelyOrosz putting the "Pragmatic" in "Pragmatic Engineer":

[Gergely Orosz @GergelyOrosz]

On the whole “just use loops”

Outside of the increasingly few people who

1) have unlimited AI token budgets

2) feel like prompting agents are holding them back (usually thanks to no #1)

I don’t think many have a use case for them. I’m more than content prompting (esp w #1!)

[Lester Kim @LesterCKim]

@TTrimoreau My own business, focusing largely on customer discovery and market research while I cover living expenses.

[Lester Kim @LesterCKim]

@DanielSmidstrup Depends on what you mean by "check". Read? No. Test? Yes.

[Lester Kim @LesterCKim]

@NWExplained Customers don't care what you call yourself. They care that you solve their problem.

[Lester Kim @LesterCKim]

@Tech_girlll Using AI to generate code without tests. (Definition inspired by @mfeathers's definition of legacy code)

[Lester Kim @LesterCKim]

I asked @hotuan at @xfund in 2024, and he said "lack of focus". I agree. As a startup, the one advantage you should have is the ability to focus all your time and effort into one area. However, most startups end up spreading their limited resources too thinly across multiple projects. @AlexHormozi often preaches the same. Focus is a superpower.

[Lester Kim @LesterCKim]

Software/AI agencies waste your time and money. If you want to use them effectively to build custom applications, do the following:

1. Have them contribute using accounts that you control (e.g. GitHub, AWS, Claude).
2. Have them write tests that document the expected behavior of the system.
3. Have them give daily updates that describe what they did the previous, what they plan to do today, and what they are blocked by.
4. Have them deploy using infrastructure as code (IaC) tools like Terraform.
5. Have them deploy something to production hosted on the web so that you can see the progress they make. This should be done as early as possible BEFORE the first product feature is done.

As long as you do this, if the contractors ever disappear (which happens) or they don't work with you for whatever reason, you can bring another engineer on without much hassle. This should be done for every new project with a development shop.

[Lester Kim @LesterCKim]

@AliGrids Infinite Claude because I can use it to build skills to invest in myself. That’s worth far more than Bitcoin and I’ve worked mostly in crypto in my career.

[Lester Kim @LesterCKim]

@babayagatwt Not now but maybe years down the line with inflation and an older Gen Z with a mix of deep engineering skills and sales ability to operate in a highly regulated B2B market.

[Lester Kim @LesterCKim]

@vivoplt For a web app, a walking skeleton that spans both

[Lester Kim @LesterCKim]

@24hoursMMA A decent wrestler will keep his head up when going for a single or double. In MMA, he will set up the takedown to clear the arms. The proper thing to do is to learn how to wrestle so you can defend the takedown.

[Lester Kim @LesterCKim]

I disagree. The way you actually generate viable business ideas generally require you to dive deeply into a business domain with many conversations within a specific sub-niche to identify that ONE common pain point that people are willing to pay money for you to solve. AI can certainly help with certain parts, but AI is not going to do the difficult, real impactful work of customer discovery to find the best ideas.

[Lester Kim @LesterCKim]

In terms of working in the industry in startups, find a narrow niche of prospects for whom you solve a single painful problem with your software. You do this through customer discovery so that you identify latent demand for generating good product ideas. This is the MOST IMPORTANT thing that matters when starting out.

[Lester Kim @LesterCKim]

@khushiirl These days, yes. It wasn't always this way. The era of heavy frontend frameworks where state entered the client-side made frontend VERY complicated. This is why most web apps should start off with SSR (server-side rendering) where state is stored server-side as much as possible.

[Lester Kim @LesterCKim]

It's not an MVP if you don't already have a very specific customer base for whom you solve a single, painful problem they're willing to pay money for your product to solve. If you're doing this right with no audience, then yo u should already have a professional network within an industry. If you don't even have that, then you need to focus on customer discovery WAY before even thinking about an MVP.

[Lester Kim @LesterCKim]

@1Umairshaikh Launch right now and move on. Most likely, it'll gain no traction. Better you learn that now. You should have prospects waiting for your product BEFORE you launch. As much as possible, establish your first trial users who signed up AND THEN build the MVP.

[Lester Kim @LesterCKim]

@bjeansonne @paulg It depends on which college. The rich parents I know encourage all kids to get into the best colleges they can. That said, if the kid is weak academically and disinterested in college, he is better off working right away or joining an apprenticeship.

[Lester Kim @LesterCKim]

@paulg For most, this is true. If you're weak academically and have no interest in school, you're better off not going to college. However, if you get into an elite university (e.g. Harvard, MIT, Stanford), you're better off attending because of the opportunities it'll open up for you.

[Lester Kim @LesterCKim]

The two core skills of any tech leader are:

1. Listening
2. Asking questions to uncover key information

Both boil down to genuine curiosity, not just about technology and product ideas, but primarily about people: customers, users, and teammates.

The most senior leaders should be superb at these skills because they will inspire (often subconsciously) the rest of the company to operate accordingly.

I developed these skills from doing theater, especially improv. I highly recommend taking a class. You'll probably learn more through that than a typical professional development course AND have more fun!

If you're in the NYC metro area, I recommend The @magnettheater. I learned a lot from every teacher I studied with there: Armando Diaz, Rick Andrews, @KornfeldLouis, @nickkanellis, @Micnen, Michael Lutton