<G /><Gaurab />
  • ⌘+K

    Command Palette

    Search for a command to run...

  • projects
  • devLogs
  • resume

<Gaurab />

pages

projectsdevLogsresume

connect

GitHubLinkedInScholarEmail

By Gaurab Chhetri on August 30, 2025

What Four Hackathons Taught Me About AI, Teamwork, and Sleep Deprivation

What Four Hackathons Taught Me About AI, Teamwork, and Sleep Deprivation

When I moved to the US in 2024, I promised myself I’d jump into the hackathon scene. Not because I needed more late nights in front of a laptop (though that’s guaranteed), but because hackathons compress months of learning, collaboration, and experimentation into a single weekend.

Since then, I’ve participated in four hackathons, and somehow managed to win three of them. Along the way, I’ve built everything from an AI-powered medical assistant to an emergency response system. But more importantly, I’ve walked away with lessons that go far beyond code.

Lesson 1: AI is only powerful when grounded in real problems

At almost every hackathon, you see teams rushing to plug GPT into something flashy. The real winners, though, are the ones who ground AI in human needs.

  • With MedGuide AI 1, we weren’t trying to show off prompt engineering tricks. We wanted to make medical documents less intimidating. That human-first focus is what made the project resonate with both judges and users.
  • With ResQMe 2, the AI wasn’t the star, it was the bridge. It triaged SOS messages into “critical,” “moderate,” or “minor,” giving responders a head start instead of a black-box chatbot.

I learned that the best AI projects don’t chase what’s possible, they solve what’s painful.

Lesson 2: Teamwork > Tech Stack

Hackathons throw together people with different skill levels, backgrounds, and even sleep schedules. In my teams, I had members new to Python, teammates more comfortable in MERN, and others who were primarily designers.

The real breakthrough came when we stopped worrying about who knew what, and instead focused on dividing tasks in a way that played to everyone’s strengths.

  • One person owned backend APIs.
  • Another handled front-end polish.
  • Someone else set up deployment pipelines.
  • And we made time for knowledge sharing, even in the middle of the chaos.

The result? Projects that didn’t just work, but worked well enough to impress judges in under 48 hours.

Lesson 3: Sleep deprivation is real, but so is adrenaline

Hackathons have this strange rhythm: you hit a wall at 2 AM, swear you’re done, then suddenly get a second wind at 3 AM and push through until sunrise.

At RiverHacks, my team and I literally pulled an all-nighter to finish ResQMe. We were delirious by the end, but that’s also when we found the “one more feature” that took the project over the top.

I don’t recommend making no-sleep weekends a lifestyle, but I’ve realized that constraints spark creativity. When time and energy are short, you cut the fluff and focus on what truly matters.

Lesson 4: Hackathons are accelerators for growth

In less than a year, hackathons have given me:

  • Hands-on experience with LangGraph, RAG pipelines, and LLM orchestration.
  • Confidence in teamwork, even with new people.
  • Stories worth sharing in interviews and blogs (like this one).
  • A reminder that building things fast can also mean building things that matter.

More than prizes or cash, the real reward has been the growth; technical, personal, and professional.

Closing Thoughts

Hackathons aren’t just about showing what you can build in 24 hours. They’re about exploring what you want to build, alongside others who share the same drive.

Four hackathons in, I’m grateful for the wins, the losses, the sleep deprivation, and the friendships made at 3 AM debugging sessions. If anything, they’ve taught me this:

👉 AI is powerful, but teamwork makes it impactful. Constraints hurt, but they also clarify. And building for people, not prizes, is the ultimate win.

Here’s to many more weekends of code, coffee, and maybe; just maybe; a little more sleep.

References

Footnotes

  1. https://github.com/gauravfs-14/medguide-ai ↩

  2. https://github.com/gauravfs-14/ResQMe ↩

github ·linkedin ·scholar ·email ·computenepal

© 2025 Gaurab Chhetri. All rights reserved. ·llms.txt·sitemap.xml