Wow you write well! I really like your point about being proactive finding work to do. This definitely makes life easier, as you're no longer at the mercy of your Jira board

AI/ML Eng @ Series C startup

Elliot Kang

In my enthusiasm, it looks like I missed a few points 😅 Some things I notice about really, really good engineers is that

#### Breaking Down Work
They're just so good at breaking things into atomic tasks. If there's a whole new workflow, they break it down into as many trivial tasks as possible and knock them out.

Some of Alex's videos on "how real system design/feature planning is done" are super helpful. There's like 3-5 of these at least somewhere.

#### Getting the right people involved
If there's a feature that involves approval from 3+ people, I've seen them just get ahead of the work and gather requirements super efficiently.

TLDR; good engineers are really good at **doing the preparation before coding** so they have to write 70% less code when the implementation actually begins. WLB does exist, but I think you'll have to fight hard for it.(being efficient, constantly improving, picking the right companies/teams)

Oh man. So. I said yes to everything when I first started being a SWE 6 months ago. It was a huge mistake. I was trying to do too many things at once. Ultimately, I took a huge reputation hit and it took months to recover.

So after reading your post several times, I'm getting the impression that you're putting too much pressure on yourself. The expectation of any new employee is to **get the most important work done first**. That's pretty much it. Then you add stuff like code review, refactoring, etc.

I wish I knew a few of these non-ENG related (but even more important) things before I got into tech... 😴

#### Having Boundaries
If your teammates ask you for help and you're swamped with work, just say no. They'll understand, almost any reasonable person will. This is really important because if you never say no, people may just keep asking you to do things.

#### Stretching Deadlines (in moderation)
You're never going to have 100% days all the time. So if there's a seemingly small project but there's a lengthy review process, it's good to give yourself 20-30% breathing room to put less pressure on yourself.

#### Not trying to learn everything at once
If you try to learn everything at once, you'll learn nothing. How are you supposed to know what the code reviewers on your team look for, before you've even landed any code? 

It takes time to meld your coding style with the team's. It takes time to give consistent and thoughtful code reviews to your coworkers. It takes time to add value via ad-hoc tasks.

^side note: I think it's actually harder to add value with ad-hoc tasks than your main work. They are prone to be poorly scoped and not prioritized properly, so it could be a huge time sink to work on these if you're new.

#### Final Thoughts
Enjoy the grind! Make sure to have some fun. Touch some grass. Blow some money on concert tickets or something. As a highly paid SWE, it's your turn to stimulate the economy 💸

Entry-Level Software Engineer at Seed Startup

There should be resources online on the expected equity percentage. That being said, I'd give the "equity" at least a 70% haircut when you tack it onto base as "TC".

Talking to the engineers would definitely help.

Oh yeah, final thing. If it's a Series A startup, ask what the milestones are to hit Series B and where the team is at in regards to that. If the startup is at $0.1 million ARR and it needs to hit $20 million, that's just an uphill battle.

Your concerns are valid. There is only so much you can find out about a startup during the interview process. And it's unlikely they'll tell you about any of the real problems, so you have a ton of guess work to do. If they do tell you about real struggles, that's a pretty good green flag.

Founders tend to say whatever they need to in order to get early employees to join. Sadly, you don't have Glassdoor at this stage to do some due diligence.

### Common Red Flags

#### Shiny Demo

Never, ever believe any demo that they give you (especially if it's an AI demo). It's most likely fake, unless they let you use the tool yourself.

#### Equity-Heavy offer

Early stage startup valuations are rather fuzzy. And if they have stacked SAFE contracts, your equity can get diluted really fast. Don't let startup founders give you this "but your equity is worth $5 million right now according to my personal valuation...", that's just a lie.

#### High Turnover

Look up both current and former employees on Linkedin. If the turnover is really high, chances are you're looking at some unstable personalities steering the ship.

#### Nobody Leveling Up at the Startup

Ask how people get promoted or significant increases in comp. If nobody has gotten increased responsibilities/comp in the past 6 months/year, that means nobody is given the opportunity to grow.

#### Low-No revenue/High Churn

Money talks. If a startup isn't getting traction, it's not getting traction.

### TLDR;

Sadly, I can't name all the red flags. There are many unscrupulous founders who will bullshit you to the moon and back. But by asking hard questions during the interview process, hopefully you'll be able to filter out most of the grifters.

## Addressing Resume Gaps
You're not alone, I have a gap on my resume too. Did a Master's in CS and then dropped out to pursue this GenAI thing. So yeah, I get it.

### The Market and Resume Gaps
I got an offer in April this year, so I've experienced some of this firsthand. Given how hard the tech market has been since last year, I doubt too many recruiters will throw out your CV because of a resume gap. And if they do, they don't deserve you anyways (as long as you've been doing something to progress your goals)

### Attitude During Hard Times
I think the more foundational thing is that you don't give up. Everyone gets stuck at some point, whether it's not getting into a top tier school, hating their first job, or not getting promoted at their first job for several years. I even had friends get into medical school after 4 years of trying. 

If you *really* want to do this and don't see yourself doing anything else (at least in the 3-5 year timeframe)... just stick with it. 



It seems like you're wondering what creating a startup will be like. 

Here's a quick way to see what it's like, without quitting your job and making your partner angry:

Just go live a day in the life with a startup founder (preferably several to get a representative view).

If you work remote 1-2 days a week, you can go to an AI event or hackathon and look for startup founders to cowork with. Then you can pick their brain over lunch/dinner after coworking at a cafe or something. (they might even invite you to a founder hotbed like SHACK15, which has even more founders)

Early stage founders tend to have much more insight into things like:
- what the funding climate is actually like in your region
- how to deal with the stress of being an early stage founder
- hiring the first few people, etc

This is important because in 2024, the ground truth and what mainstream news reports about tech diverges significantly at times.

Anyway, hope this helps. I haven't started a startup yet, but it gives me peace to at least know what I'd be signing up for if I did.



I think most of those startups are definitely later stage, which are much safer. It's just that there is the constant risk of being no longer seen as that "breakout startup". For example, OpenAI releasing a killer search engine would hurt Perplexity.

Mistral this time last year would have been considered a breakout startup :/

I'd just be wary of any of the generative AI startups on that list, some of them have serious customer churn (so they wouldn't be considered breakout startups) and are having trouble raising a next funding round.


Well. To give you some hope, I was jamming Cursor AI with a bunch of questions for 20 minutes about how the opentelemetry library worked. What I really needed was a 3-line change, which I would have known had I worked with the library before.

Cursor definitely helps. But I've noticed that the larger the codebase, the worse Cursor does. For max accuracy, I've noticed that it does best on knowable work that has lots of documentation.

Things Cursor is bad at:
- refactoring code where the bugs/edge cases are unknown
- writing code that based on other internal code

btw I'm not sure if o1 actually helps, it's a little too expensive for my taste. I just stick to o1-mini for now. But hey, I'll take the free productivity boost for $20/month

Well if you can get into OpenAI or Anthropic, absolutely. But if you're part of the 99.99% that can't get in, it's okay. I knew someone who got into Midjourney and he had to be so so proactive to even get noticed.

Taro can give you all the resources in the world, but the harsh reality is that there is very little outside information on private companies. So if I were in your shoes, I'd try for big tech simply because there is much less unnecessary risk.

You never know if your startup
- is about to hit a major down round
- doesn't have product-market fit
- has 2 weeks of payroll left
- even has a good engineering culture

Especially if you're looking for a quick cashout, founders will totally sell you that dream to get you to join.

Some harsh truths:
- Work-life balance is worse for sure, unlike FAANG there are tighter deadlines because you should have shipped yesterday
- TC doesn't mean anything for you, it's only for your employer to rationalize paying you less
- Chances are, nobody has heard of your startup. And 99% of the time they never will.

If you think this would never happen to a break-out startup, it can at any moment. Meta is absolutely killing every other open-source LLM startup, including Mistral (which had the world's best cost/performance LLM for a few months).

Now, reasons to work at a startup despite all the instability:
- you get to ship fast, features take 1 week instead of 4 months
- you can impact the startup's success
- you get to work on a variety of things (infra, features, maybe even some ML)
- your coworkers are likely to enjoy their job
- unlike big tech, they won't sunset your project after you've sweat blood finishing it

Ultimately, nothing from the startup will likely remain. The equity can be worth pennies, nobody will remember how many all-nighters you pulled, and people don't care about the startups that never made it. What remains is your skills and experiences.

Well, at least that's what I tell myself so I can sleep at night

I was in your same position exactly a year ago, so this hits really close to home. It was so hard to get interviews that even with spam applying, I only got 2 callbacks in a month. Then I started contributing to open-source, and things started looking up. Here's some pointers:

\*\*For those who are later in their career, I think these principles still apply. The impact of doing this earlier in your career has more of a 0-1 effect than the 1-5 effect someone later in their career would see IMO

### **Pick A Niche**

So there's a deluge of MSCS graduates who want to work in AI. You need to focus on a single direction. There's several routes you can take:

- Model Serving
- Classic ML model development (collaborative filtering, etc)
- Large ML models (LLMs, stable diffusion, etc)
- Fine-tuning
- Generative AI tools
- ML infra tools like Flyte/Airflow
- Classic ML model dev and Large Model work typically require top conference papers to be in the top 1%

Before you dive into any route, make sure you do some due diligence so that you know roughly what it takes to be the top 1% in anything. Then decide if you're willing to put in the work.

### **Pick The Project Size**

Then there's size of the open source project. Small libraries/frameworks are too unknown to get you any real attention with recruiters. Large libraries/frameworks will take 1-2 years to get on the PMC.

Small projects are things like util libraries. Large projects are established frameworks like PyTorch or Kubernetes.

The "goldilocks zone" IMO is medium-size projects that are rapidly gaining traction. They have enough things on their roadmap that'll add immediate value and they should be the most welcoming of new contributors.

### **Making Core Contributions**

Start with something small. No sane person would ever let a stranger add changes to their core codebase. Start with something small like a cherry-pick commit or an integration.

Then it's time to align your contributions to the actual product roadmap. Attend contributor meetings. Talk to other contributors on Slack/Discord. Get involved.

As you contribute, you'll see yourself go up the "top contributors" list. You'll see yourself writing better PRs, writing cleaner code, and working on more impactful features.

### **Spinning Open-source Into Opportunities**

Getting to the top 10 or top 5 contributors for a medium-size library doesn't take that long. Even in the span of 1-2 months, you can get to Top 5 if you put in full-time effort.

Now you have some awesome résumé bullet points:

- Built core features: X, Y, Z, etc
- Top 5 contributor for Github project with 20k stars and 500K downloads/month

Now go! Find a project and contribute. You'll learn a lot + be helping out the community. I think that you can only re-write your resume so many times before you need to make a fundamental change. Making open-source contributions can help you make that fundamental change to your profile.