My coworkers and I have noticed those who are top players in videos games tend to be cracked engineers.
If you’re one of them, I want to hire you. Come build in SF with me, we’re building something incredible at @vitalize_care
My coworkers and I have noticed those who are top players in videos games tend to be cracked engineers.
If you’re one of them, I want to hire you. Come build in SF with me, we’re building something incredible at @vitalize_care
Good companies don't hand their hardest, most critical projects to engineers that just walked in the door with a history of leaving early
If you want to solve those problems (and add them to your resume), you've got to stick around and earn it
@_arnavchauhan Looking back I think it’s because we had date strings like 10/25/24 being passed around, and parsing all the configs it can handle, it was faster to do this fix the root problems 😭
In healthcare ops, infra performance directly impacts how fast hospitals can respond to staffing gaps.
Supabase was good for early stage. It helped us move fast. But as we scaled across major health systems, we hit limits around connection pooling and unpredictable latency under load.
We just migrated Vitalize from Supabase to @PlanetScale Postgres:
Zero downtime.
400GB.
150M+ rows.
Now:
P95 latency: 2ms
200GB JSONB reads: ~1.2s, down from ~50s
Fewer resources, far more predictable performance
https://t.co/gdwW6XXHVI
In healthcare ops, infra performance directly impacts how fast hospitals can respond to staffing gaps.
Supabase was good for early stage. It helped us move fast. But as we scaled across major health systems, we hit limits around connection pooling and unpredictable latency under load.
We just migrated Vitalize from Supabase to @PlanetScale Postgres:
Zero downtime.
400GB.
150M+ rows.
Now:
P95 latency: 2ms
200GB JSONB reads: ~1.2s, down from ~50s
Fewer resources, far more predictable performance
https://t.co/gdwW6XXHVI
Even with lots of human in the loop intervention, I feel like my code quality has gone down. Things are a lot more verbose, the if statements can run deep. Need to figure out how to reign those in.
On the bright side, I've never added as many tests in my life.
This month I've been really trying out LLMs. Coding like this feels magical, but also scary. Got burned from trusting the LLMs too much earlier on in the month, but feel like I have good commands now and plenty of human in the loop intervention to keep it on a straight path.
Any tips on limiting context when cursor is fixing tests? I find it easily eats up a lot of my tokens, especially in bigger test files with a lot of concurrency.
@bentlegen@modemdev Not related to the video -- what's the sound track? I got distracted while this video was playing, and was getting into the zone with it.
@imax153 💯. So excited for cursor & graphite. I dream of the day I can leave comments directly in the review mode vs trying to coordinate it in the chat
I have a growing opinion that being a good software engineer is going to be less and less about what you can build, but more of how you are able to take preexisting code, navigate context of it, and continue to iterate on it while making it more reliable.
@wyatt_sg I agree! Knowing when/how is very important.
Can’t keep building on terrible foundations, but being able to iterate on it that sets you up better for the future while continuing to provide value is a huge skill.
I think we’re starting to see the cons of how cheap code is now.
Software we all use seems to be shipping faster than ever, but introducing so many bugs that break critical workflows.
Excited to see the solutions people build to help automate testing more and catch bugs
It’s very easy to build now with AI.
It’s very easy to look at the code, and want to change everything about it.
It’s much harder to continue to ship wins, produce software that stays reliable, and ship wins for everyone on your team.