Manav Gandhi
@manavgandhi27.bsky.social
Full Stack Software Engineer building scalable, production-grade systems.
Experience across frontend, backend, and system design.
I value clean architecture, correctness, and solving real business problems.
Experience across frontend, backend, and system design.
I value clean architecture, correctness, and solving real business problems.
Your tests pass.
Code review approves.
Staging looks perfect.
Production breaks anyway.
Here's what tests miss 🧵
Code review approves.
Staging looks perfect.
Production breaks anyway.
Here's what tests miss 🧵
February 8, 2026 at 4:32 AM
Your tests pass.
Code review approves.
Staging looks perfect.
Production breaks anyway.
Here's what tests miss 🧵
Code review approves.
Staging looks perfect.
Production breaks anyway.
Here's what tests miss 🧵
Want to be the best developer in 1 year?
Here are 5 books that changed how I code, think, and build.
Not theory. Real, applicable stuff 🧵👇
Here are 5 books that changed how I code, think, and build.
Not theory. Real, applicable stuff 🧵👇
February 8, 2026 at 3:52 AM
Want to be the best developer in 1 year?
Here are 5 books that changed how I code, think, and build.
Not theory. Real, applicable stuff 🧵👇
Here are 5 books that changed how I code, think, and build.
Not theory. Real, applicable stuff 🧵👇
Two types of developers:
Type A: Perfects tutorials for 6 months, never ships
Type B: Ships broken projects, learns by iterating
Which one are you? 👇
Type A: Perfects tutorials for 6 months, never ships
Type B: Ships broken projects, learns by iterating
Which one are you? 👇
February 8, 2026 at 3:51 AM
Two types of developers:
Type A: Perfects tutorials for 6 months, never ships
Type B: Ships broken projects, learns by iterating
Which one are you? 👇
Type A: Perfects tutorials for 6 months, never ships
Type B: Ships broken projects, learns by iterating
Which one are you? 👇
Are your fallbacks actually working or are you just hoping?
I stopped assuming and started breaking things on purpose.
Here's what I learned 🧵
I stopped assuming and started breaking things on purpose.
Here's what I learned 🧵
February 7, 2026 at 12:54 PM
Are your fallbacks actually working or are you just hoping?
I stopped assuming and started breaking things on purpose.
Here's what I learned 🧵
I stopped assuming and started breaking things on purpose.
Here's what I learned 🧵
Your system makes 1000 decisions per day.
Can you explain any of them?
"Why didn't my PR get reviewed?" should take 10 seconds to answer, not 10 minutes of log archaeology.
Here's the fix.
Can you explain any of them?
"Why didn't my PR get reviewed?" should take 10 seconds to answer, not 10 minutes of log archaeology.
Here's the fix.
February 6, 2026 at 9:45 AM
Your system makes 1000 decisions per day.
Can you explain any of them?
"Why didn't my PR get reviewed?" should take 10 seconds to answer, not 10 minutes of log archaeology.
Here's the fix.
Can you explain any of them?
"Why didn't my PR get reviewed?" should take 10 seconds to answer, not 10 minutes of log archaeology.
Here's the fix.
12 months ago, I wouldn't have had the guts to post my code publicly every day.
Now, I don't just do it — I teach others how to build in public.
Here's what changed 🧵👇
Now, I don't just do it — I teach others how to build in public.
Here's what changed 🧵👇
February 6, 2026 at 1:12 AM
12 months ago, I wouldn't have had the guts to post my code publicly every day.
Now, I don't just do it — I teach others how to build in public.
Here's what changed 🧵👇
Now, I don't just do it — I teach others how to build in public.
Here's what changed 🧵👇
I added a second server to my app.
Everything broke instantly.
Duplicate notifications. Exceeded rate limits. Servers fighting each other.
This is distributed systems. Most learn it the hard way in production.
Here's the fix.
Everything broke instantly.
Duplicate notifications. Exceeded rate limits. Servers fighting each other.
This is distributed systems. Most learn it the hard way in production.
Here's the fix.
February 5, 2026 at 9:30 AM
I added a second server to my app.
Everything broke instantly.
Duplicate notifications. Exceeded rate limits. Servers fighting each other.
This is distributed systems. Most learn it the hard way in production.
Here's the fix.
Everything broke instantly.
Duplicate notifications. Exceeded rate limits. Servers fighting each other.
This is distributed systems. Most learn it the hard way in production.
Here's the fix.
Here are 7 tools I use every week as a full-stack developer — and why I love them 🧵👇
February 5, 2026 at 2:26 AM
Here are 7 tools I use every week as a full-stack developer — and why I love them 🧵👇
What breaks when you scale from 1 server to 2?
Everything.
After 9 days building MergeSense, I asked this uncomfortable question. The answer forced me to rethink how I design systems.
Everything.
After 9 days building MergeSense, I asked this uncomfortable question. The answer forced me to rethink how I design systems.
February 4, 2026 at 9:45 AM
What breaks when you scale from 1 server to 2?
Everything.
After 9 days building MergeSense, I asked this uncomfortable question. The answer forced me to rethink how I design systems.
Everything.
After 9 days building MergeSense, I asked this uncomfortable question. The answer forced me to rethink how I design systems.
Here's my step-by-step process for going from 0 to deployed full-stack app in 30 days.If you're trying to ship your first real project (not a tutorial clone), here's what worked for me 🧵👇
February 4, 2026 at 2:18 AM
Here's my step-by-step process for going from 0 to deployed full-stack app in 30 days.If you're trying to ship your first real project (not a tutorial clone), here's what worked for me 🧵👇
Your AI tool will crash from its own success.
I built MergeSense for 8 days. Then realized: I never tested what happens when 50 PRs arrive simultaneously.
Ran the numbers. 50 concurrent requests = 1GB memory. My server: 512MB.
Crash.
I built MergeSense for 8 days. Then realized: I never tested what happens when 50 PRs arrive simultaneously.
Ran the numbers. 50 concurrent requests = 1GB memory. My server: 512MB.
Crash.
February 3, 2026 at 10:30 AM
Your AI tool will crash from its own success.
I built MergeSense for 8 days. Then realized: I never tested what happens when 50 PRs arrive simultaneously.
Ran the numbers. 50 concurrent requests = 1GB memory. My server: 512MB.
Crash.
I built MergeSense for 8 days. Then realized: I never tested what happens when 50 PRs arrive simultaneously.
Ran the numbers. 50 concurrent requests = 1GB memory. My server: 512MB.
Crash.
If you're struggling with staying consistent while coding full-time, try this instead.Capture, don't create. Here's what I mean 🧵👇
February 3, 2026 at 2:29 AM
If you're struggling with staying consistent while coding full-time, try this instead.Capture, don't create. Here's what I mean 🧵👇
Every founder asks the same question about AI tools.
"How much will this actually cost me?"
Most companies can't answer with data.
I just built the system that does—in 290 lines of code, no database required.
"How much will this actually cost me?"
Most companies can't answer with data.
I just built the system that does—in 290 lines of code, no database required.
February 2, 2026 at 10:30 AM
Every founder asks the same question about AI tools.
"How much will this actually cost me?"
Most companies can't answer with data.
I just built the system that does—in 290 lines of code, no database required.
"How much will this actually cost me?"
Most companies can't answer with data.
I just built the system that does—in 290 lines of code, no database required.
New month, new energy. Here's what I'm focusing on in February 🧵👇
February 2, 2026 at 2:29 AM
New month, new energy. Here's what I'm focusing on in February 🧵👇
Day 7: I didn't ship features. I shipped accountability.
No AI improvements. No performance gains.
Just the boring work that lets you answer "why?" when production breaks at 3 AM.
This is what separates demos from systems you actually own. 🧵👇
No AI improvements. No performance gains.
Just the boring work that lets you answer "why?" when production breaks at 3 AM.
This is what separates demos from systems you actually own. 🧵👇
February 1, 2026 at 11:30 AM
Day 7: I didn't ship features. I shipped accountability.
No AI improvements. No performance gains.
Just the boring work that lets you answer "why?" when production breaks at 3 AM.
This is what separates demos from systems you actually own. 🧵👇
No AI improvements. No performance gains.
Just the boring work that lets you answer "why?" when production breaks at 3 AM.
This is what separates demos from systems you actually own. 🧵👇
Reposted by Manav Gandhi
I recently experienced a surreal moment at work. We had a project slipping down the backlog for weeks because the estimated effort was too high. We spent days in meetings debating the approach.
Then a coworker just... did it. Used AI to scaffold the whole thing in one day.
Then a coworker just... did it. Used AI to scaffold the whole thing in one day.
January 30, 2026 at 8:23 AM
I recently experienced a surreal moment at work. We had a project slipping down the backlog for weeks because the estimated effort was too high. We spent days in meetings debating the approach.
Then a coworker just... did it. Used AI to scaffold the whole thing in one day.
Then a coworker just... did it. Used AI to scaffold the whole thing in one day.
Want to get better at building in public, shipping consistently, and growing as a developer?
Follow these 5 devs doing it right 🧵👇
Follow these 5 devs doing it right 🧵👇
February 1, 2026 at 2:40 AM
Want to get better at building in public, shipping consistently, and growing as a developer?
Follow these 5 devs doing it right 🧵👇
Follow these 5 devs doing it right 🧵👇
Want to get better at building in public, shipping consistently, and growing as a developer?
Follow these 5 devs doing it right 🧵👇
Follow these 5 devs doing it right 🧵👇
February 1, 2026 at 2:40 AM
Want to get better at building in public, shipping consistently, and growing as a developer?
Follow these 5 devs doing it right 🧵👇
Follow these 5 devs doing it right 🧵👇
Day 6: I integrated Claude AI into MergeSense.
The easy part: 20 minutes
The hard part: 6 hours
The easy part was making an API call.
The hard part was making it work when the API times out, returns garbage, or the network dies.
Most build-in-public posts show the demo. 🧵👇
The easy part: 20 minutes
The hard part: 6 hours
The easy part was making an API call.
The hard part was making it work when the API times out, returns garbage, or the network dies.
Most build-in-public posts show the demo. 🧵👇
January 31, 2026 at 11:30 AM
Day 6: I integrated Claude AI into MergeSense.
The easy part: 20 minutes
The hard part: 6 hours
The easy part was making an API call.
The hard part was making it work when the API times out, returns garbage, or the network dies.
Most build-in-public posts show the demo. 🧵👇
The easy part: 20 minutes
The hard part: 6 hours
The easy part was making an API call.
The hard part was making it work when the API times out, returns garbage, or the network dies.
Most build-in-public posts show the demo. 🧵👇
One line I keep coming back to is: "Done is better than perfect."
Simple. Overused. But it changed everything for me 🧵👇
Simple. Overused. But it changed everything for me 🧵👇
January 31, 2026 at 6:03 AM
One line I keep coming back to is: "Done is better than perfect."
Simple. Overused. But it changed everything for me 🧵👇
Simple. Overused. But it changed everything for me 🧵👇
Day 5 of building MergeSense: I spent 6 hours refactoring code no one will ever see. No features added. Just replacing boolean flags with production-grade risk detection.
January 31, 2026 at 6:03 AM
Day 5 of building MergeSense: I spent 6 hours refactoring code no one will ever see. No features added. Just replacing boolean flags with production-grade risk detection.
Fill in the blank: I wish more people knew _____ about building a career in tech.
My answer: You don't need permission to start.
What's yours? 👇
My answer: You don't need permission to start.
What's yours? 👇
January 31, 2026 at 6:02 AM
Fill in the blank: I wish more people knew _____ about building a career in tech.
My answer: You don't need permission to start.
What's yours? 👇
My answer: You don't need permission to start.
What's yours? 👇
The best code reviewers know what to ignore. Today I documented what MergeSense will never comment on. This is a permanent exclusion list—things the system will detect and deliberately ignore.
January 31, 2026 at 6:02 AM
The best code reviewers know what to ignore. Today I documented what MergeSense will never comment on. This is a permanent exclusion list—things the system will detect and deliberately ignore.
Quick check-in on Day 11 of Creator Camp.
No big wins today. Just progress. And that's enough 🧵👇
No big wins today. Just progress. And that's enough 🧵👇
January 31, 2026 at 6:02 AM
Quick check-in on Day 11 of Creator Camp.
No big wins today. Just progress. And that's enough 🧵👇
No big wins today. Just progress. And that's enough 🧵👇
I spent today architecting an AI code reviewer.
Wrote zero code.
Made all the important decisions.
Because in software, architecture is choosing what not to build.
Wrote zero code.
Made all the important decisions.
Because in software, architecture is choosing what not to build.
January 31, 2026 at 6:01 AM
I spent today architecting an AI code reviewer.
Wrote zero code.
Made all the important decisions.
Because in software, architecture is choosing what not to build.
Wrote zero code.
Made all the important decisions.
Because in software, architecture is choosing what not to build.