I saw the LinkedIn post, which was the story of my life.
None of this is LLM written, I hope this is worth your time :)The only thing I screen for when hiring engineers is how fast they learn things they don't already know.
I like engineers who don't see boundaries between what's their job and what isn't. Backend engineer who can unblock themselves on frontend.
That willingness to stretch into unfamiliar territory is worth more to me than ten years of deep expertise in one narrow stack.
Everyone says they're a fast learner. If someone has that trait they'll have a track record
I went from not knowing the difference between client/server to:
| readmaxxing | reading O'Reilly Java The Complete Reference (1k+ pages) end to end to learn to code. First of many books. |
| formal backend | becoming a reliable backend dev at an enterprise that served thousands of users, millions of endpoints (we had on-premise customers, and used "make no mistakes" long before claude) |
| formal backend++ | leading teams at the enterprise co, 10k+ dev org; being the point of contact for security reviews and guidance |
| security++ | wrote our 0-day playbook, reached consensus on DRIs (ah, enterprise), presented security findings to 500+ devs |
| back to 0 | leaving all my social capital at the bigCo to join a then-nameless startup as founding engg |
| formal tech++ | do literally everything at the startup I had teams in bigCo for - infra, backend, deployment, sec, etc |
| formal non-tech | all the while talking to sales/support tech to make their lives easier with bots, automation, browser extensions |
| desktop apps | building cross-platform desktop apps (ElectronJS), native desktop apps (MacOS) on the side |
| devops | volunteering to do devOps on AWS for random people on Discord during covid (a watch party app. I had never used AWS before) |
| mobile apps | attempting to build mobile apps (Android, failed for lack of research before implementation) |
| ai ai ai | literally staying on the bleeding edge of AI so I have a dependable workflow for myself (I strongly believe everyone needs to shape their own tools when it comes to AI adoption) |
| whatsapp bot | building an AI assistant as a TUI app, and then a whatsapp-first product till Meta blocked my primary account :( |
| framework mania | building random apps with Svelte, React, AstroJS just to try them out |
| homeserver phase | setting up my homeserver to run claude code on my phone + set up backups with Immich etc |
| browser extension | releasing a Firefox extension to preview fonts live on websites bc I was so obsessed with fonts (just look at my website) |
| ai pilled | building a real-time security monitoring solution with Golang (I don't know Go, but it was the best tool for the job) |
| beware salesforce | building a PRM for myself because I keep forgetting important events (also finds cool people on Hackernews for me) |
| hmmmmm | working at Julius AI and crushing it even though I haven't worked with a ton of data before (this one's wip) |
You name it, I've tried it. Even my websites have been hosted on Vercel, Cloudflare, Netlify, Heroku just to see how their UX differs.
I spend weeks deep in something like a maniac - everything from pre-auth RCEs on the SAML protocol to isolation levels across DBs.
But if you ask me a targeted question about them now: *clenches fist nervously, sweats profusely, laughs squeakily, chair backs up slowly*
So if I don't remember it all, what's the point?
Borrowing from Age of Empires - I intentionally aim to reduce my space of unknown unknowns.
Now when faced with a problem, I know where to look and what to look for, how to approach it, how to break it down, what to spend the most effort on, what patterns it matches, where the footguns could be, etc.
Luckily, when I know what questions to ask, LLMs are good at giving me the data!
Why Julius?
I've recently been interested in exploring data, analytics and processing.
Eg about a month ago I noticed an open source data platform for agents built by an ex-PostHog engineer, and looked into it (also got a couple of minor PRs merged into it).
This is an opportunity to help users while fielding complexity - the prospect of being able to ship to so many users, meet them where they are and eventually understand their worlds - I'm drawn like a fly to a Venus Flytrap.
I also write occassionally. Primarily to share knowledge with non-technical peers.
- The kind of company I want to be a part of (#1 on HackerNews)
- Hashing explained for beginners (top post, reddit)
- LLMs explained for beginners (top post, reddit)
- My journey
Trivia:
I always knew I wanted to achieve mastery, and read the quote "it takes 10,000 hours of practice". My first side project (<1yoe) was The Wall - every hour I spent was a "brick", and I was building my wall of mastery.
Of course, it wasn't about just building it, but building it the best way. I scoured the internet for best practices, read GitLab's handbook, etc etc
- A complete, helpful readme. No ChatGPT to autogen it for me back then, of course.
- The best OSS had discussions on Github - so I had discussions with MYSELF when I made decisions. Eg: Tags vs Categories
- Similarly, the best products seemed to plan work using project management tools. I used GitLab to plan work even though I was the only dev/user.
