just built Control Plane Notebook β production-grade, fully server-rendered Go web app for internal runbooks, incident docs, and team knowledge workflows.
A 6min video explanation ποΈ(watch it)
no React. no SPA. no JSON API layer. just Go + html/template + PostgreSQL doing exactly what they're supposed to do.
what's in it: β session auth + RBAC (member / reviewer / admin) β draft β revision β approval β publish workflow β transactional audit logging + moderation β tag-based search β all of it clean DB transactions, no shortcuts
stack: Go 1.25+, chi, pgx, sqlc, scs, clean architecture, fake stores for unit tests
built this the way I'd want to find it at work. no cowboy deploys, no "we'll add auth later." governance baked in from the start.
if your team is still living in Notion with no approval trail or sharing runbooks in Slack threads, this is the kind of thing that fixes that.
repo: https://t.co/zOAQGmBvPA
stars and feedback genuinely appreciated β especially from Go folks and anyone building internal tooling π
Went through an interview a few days ago that ended in halftime with the CEO violently closing his browser & ending the interview.
It was for an "Open Source Intern" role at an Indore-based cloud startup. The CEO himself was taking the round - someone with 12+ years of experience and heavy Kubernetes/CNPG background.
He started by asking if I had done any open source contributions before. I said no. His face changed immediately. From that point, the vibe shifted.
He started going deep into Docker, Kubernetes, and especially CNPG. I had literally started learning about CNPG only two days before the call. I managed to answer most things, but I messed up one important question - how the CNPG operator is actually written (the Go part, not the YAML).
After that he asked me to explain one of my projects. I went with RedisForge. I was doing okay explaining the structure until we reached a specific part in bloom.go - the Bloom filter initialization with error rate and capacity, where we ignore the "already exists" error to keep it idempotent. I just blanked on explaining that piece properly.
He closed the tab violently. Interview over.
Iβm not going to lie, it stung. I had put in the effort to learn CNPG quickly and I could explain most of my own code. But that one gap was enough.
What Iβm taking from this is pretty clear though.
Having projects is one thing. Being able to explain the non-obvious parts - especially around error handling and idempotency - under pressure is another.
I clearly need to get better at walking through my own code like Iβm explaining it in production, not just when Iβm building it.
Also realized that some roles (especially the ones touching operators and CNCF stuff) have a much higher bar than regular backend internships. The "Open Source Intern" title wasnβt casual here.
Still processing it. These kinds of interviews expose exactly where youβre weak, even if you have decent projects. Iβll work on that.
Today marks the day I applied over 120+ internship applications via cold email/dm, linkedin, internshala, YC, wellfound etc.
landed only 1 interview.
internship is cooked in India.
Just turned down an internship that wanted:
- 1 full week of "performance evaluation"
- 1 full month of unpaid work
before even considering βΉ15k/month for 6 months.
I know how tough the market is right now. A lot of us are desperate for experience and itβs easy to feel like you have to say yes to whatever comes.
But this kind of long unpaid trial period has become way too normalized. Good companies donβt need a month of free work to figure out if someone is decent. They run proper interviews and either pay from the start or keep any unpaid period short and reasonable.
Your time has value, even as a fresher. We shouldnβt have to work for free for weeks just to prove basic competence.
If youβre in a position to walk away from offers like this, do it. Slowly, it changes what companies think they can get away with.
i recently cold emailed a lott of startups, fintechs etc for an internship
and getting noticed feels like an achievement π
we in the same boat brother @SatishC2612 π€
just built a hybrid rate limiter in go because this one question kept running through my mind:
βwhat if redis is slow or completely down β what does my rate limiter do then?β
most solutions i saw either:
- died completely (api starts throwing 5xxs)
- or became useless (no shared state across instances)
so i built one that actually solves the real problem.
it uses:
- redis token bucket as primary (true shared state)
- local in-memory token bucket as instant fallback
- a real circuit breaker (closed β open β half-open) that automatically detects redis issues, switches to local, and recovers gracefully when redis is healthy again
comes with clean http middleware (proper x-ratelimit-* headers), full test suite, and fresh benchmarks on my machine:
- token bucket β 18.36 ns/op
- local limiter β 66.11 ns/op
- orchestrator (redis hit) β 36.01 ns/op
- middleware β 445.8 ns/op
posting an article about how i made this soon..
https://t.co/RFjFQ4sUoP
time to start applying for internships.
mandatory:
> go lang
> backend specific role
> minimum 20k stipend (yeah i'm ambitious)
> remote, if not - based in pune/mumbai
internship pressure got me real good.
couldn't study, couldn't focus, couldn't communicate.
just shut myself off.
was it worth it?
only time will tell.
see y'all tomorrow.
will be posting what I learnt, did etc
the last few weeks