Naman Raj

Posted on Apr 15

We Came. We Coded. We Cleaned the Planet (and Won).

#hackathon #devjournal #machinelearning #showdev

⚙️ Set the Scene

It all started with an open problem, a crew too stubborn to quit, and 24 hours to ship something real.

We entered this national-level hackathon with a clear goal: to build something that didn’t just work but mattered. The theme was Open Innovation, and while the freedom was exciting, we knew we had to be intentional. Instead of chasing trends, we focused on a problem that’s real, relevant, and way too often ignored- waste management.

That’s how 3RVision was born. An AI-powered platform designed to help users identify household waste and figure out whether it can be reused, recycled, or resold. It’s a practical, user-focused solution for a global issue. Built with machine learning, clean UI, and an emphasis on sustainability, we aimed to make responsible living more accessible and a little bit smarter.

We weren’t trying to impress with flashy features we wanted to solve a real-world problem in a way that actually sticks.

The 24 hours were intense. Between debugging models, designing flows, and juggling backend chaos, there were moments we genuinely thought we wouldn’t make it in time. But somehow, everything clicked just enough in the final stretch. And yeah that last-minute demo that somehow actually worked? Definitely one for the memory books.

👥 The Dream Team

Every winning project needs a team that just clicks, and we had that from the jump.

There weren’t strict roles or “you do this, I do that” energy. We were all in, all the time switching gears between frontend, backend, ML, and sometimes just keeping each other sane. Everyone wore multiple hats, and somehow, it worked.

Khushi – The team lead and the brain behind the idea. She steered
the ship, worked on the ML side, and constantly pushed the UI to
look and feel better. Super picky about the frontend—and yeah, she
kept me on my toes with it 😮‍💨.
Naman (me) – Took charge of the frontend, handled components, API stuff, Authentication and also chipped in on the backend. Made sure everything worked like it should, visually and under the hood.
Arpit – The utility player who pulled off two big wins building the Chrome extension and strengthening the backend. He was the kind of dev who quietly shipped features that just worked no fuss, no fluff, just clean execution.
Shreyansh – The wild card turned MVP. He went all-in on the ML side training our custom models, fine-tuning performance, and basically doing the kind of wizardry that made the AI magic real. He joined in a little later but made a massive impact in no time

🌱 Birth of 3RVision

The idea for 3RVision came from a simple realization: there’s a ton of waste out there, and people often don’t know what to do with it. With increasing environmental concerns, we wanted to create a practical solution that helps people identify whether an item can be reused, recycled, or resold right from their phone.

It was clear that the challenge of waste management wasn’t just about knowing what could be recycled, but how to make it easy for people to make informed, eco-friendly decisions. So, we decided to tackle this with AI, combining image recognition to offer suggestions for each item whether it was fit for recycling, resale, or reuse.

There were debates along the way could we expand beyond just waste management? Should we focus on something more abstract? But in the end, we locked onto this problem because it felt practical, relevant, and scalable. Plus, we could really make a difference with it. And let’s be real, who doesn’t love building something that can make a positive impact?

🛠️ Game Time – The Hack Begins

We kicked off the hack with all the enthusiasm in the world. The morning went pretty smooth frontend coming together, models being trained, vibes were solid.

Khushi and Shreyansh were deep into training the core ML models, I was locking in the frontend, and Arpit was busy setting up the Chrome extension getting those early pieces in place so we could integrate it all later.

Around evening, the organizers hit us with a surprise jamming session and not gonna lie, it was exactly what we needed. 10/10 would jam again.

Now, back to the grind. Post-jam, things got a bit… spicy.

There were some weird bugs in the extension it wasn’t properly classifying items into degradable vs. biodegradable. Me and Arpit tag-teamed it and squashed that mess before dinner. Clean fix, good vibes

But the real pain hit when Shreyansh’s model training crashed network issues mid-process. Bro left it training during the jam, and came back to find everything wiped. Khushi, on the other side, had another model training simultaneously, so Shreyansh had to restart from scratch. 🫠

By the time dinner rolled around, everything seemed set up frontend ✅, extension ✅, ML ✅.

But the backend? Not even close.
We didn’t know it yet, but we were about to enter chaos mode.

⚡ Clutch Moments – Bounce Back Energy

Okay, so here’s where things went full “will they make it or not” mode.

By 11 PM, the backend was the only thing standing between us and a fully functional system. And it wasn’t just one integration we had to:

Connect the frontend to the ML model
Connect the frontend to the backend (built in Go)
And finally, connect the backend to the ML model (running on a Flask server)

Basically, a full-on relay race with two different servers and a lot of bugs in between.

Me and Arpit sat down thinking, “Okay, couple hours max.”
Reality check: it took us till 6 AM.

There were CORS issues, data not syncing right, weird behavior when Go tried to communicate with Flask, and a bunch of edge cases that kept popping out of nowhere. But with each successful response, we got closer.

Finally, when we saw the frontend fully talk to Go, which then made a clean call to Flask and got back a prediction it was like…
inner peace unlocked.

Watching the entire pipeline run smoothly after 6 hours of chaos?
Yeah, that moment hit different.

🎤 The Pitch – It’s Show Time

We were riding on zero sleep and infinite adrenaline.
The product? Ready.
The presentation? Not so much.

We hadn’t prepped any fancy slides or rehearsed a pitch just raw conviction and a working build. Some features didn’t make it in due to time (rip to our extra functionalities).

First round, we were called to one of the labs.
Khushi and Arpit stepped up, laptops in hand.
Khushi took the lead and bro, she snapped.
Clear, confident, and straight to the point. It wasn’t just a walkthrough—it was a proper storytelling session of what 3RVision stood for and what it could do.

We walked the judges through the features—ML predictions, waste classification, the Chrome extension in action, clean UI, and smooth backend integration.
And the live demo? Bro… it actually worked. Like, everything just clicked at the right time. No bugs, no lags. It felt good- really good.

After wrapping up, we headed to the auditorium for another jamming session.
It was just what we needed. Chill vibes, music, and a moment to breathe after the storm.

🏁 The Final Lap – Top 6, Showdown & The Win

We were just vibing in the auditorium post-demo, soaking it all in, when Khushi got a call.
We were selected in the Top 6.
For a second, we just froze. Like wait, us?

The final round was calling, and we had to show up again. No room for error now.
We prepped in silence, running through key points, refining our story this time, it wasn’t just about features. It was about vision, scalability, and the “why us?” of it all.

Back to the same lab, but this time with more judges, and a slightly heavier vibe. The pressure was real.
Still, Khushi stepped up again and killed it. This round dove deeper into the business model, revenue possibilities, and future roadmap. We highlighted the packed set of features and why this wasn't just another hackathon build, it had potential. Real, tangible potential.

There were some solid cross-questions from the panel. We held our ground and answered them all.
No blank stares. No panic. Just a team that knew what it built.

After about 15 minutes, we were told to head for Lunch.
And on the way, something wild happened
The Professor of ISE came up to us and asked for a selfie.
We were like, “Uhhh… what just happened?”
That moment? Kinda confirmed what we were all feeling inside we might just take this.

Fast forward to auditorium, final results.

Started with the usual boring speeches 😒 (no offense, Chief Guest), but all we could hear was our own heartbeat.

Third place... not us.
Bittersweet. Still hoping.

Second place... also not us.
Now we’re fully in “it’s either us or nothing” mode.

And then...

🥇 “First Position… Team Quaternary!”

Bruhhh,we exploded. Screams. Hugs. Pure disbelief.
Out of 21 solid teams, we came out on top.
The sleepless night, the endless debugging, the full-stack hustle, the caffeine-fueled grind—all of it was worth it.

🚀 That’s a Wrap

From day-one brainstorming to that final pitch, the whole journey was wild in the best way.
We built, broke, fixed, and figured things out as a team, learning more in 24 hours than we probably did in weeks.

What started as just an idea turned into something we’re genuinely proud of.
Walking away with the win? Yeah, that felt amazing.
But honestly, it’s the grind, the laughs, the chaos, and the crew that made it all worth it.

Thanks for sticking around and reading our story! 🙌
This wasn’t meant to be a technical deep-dive but more of a raw, real peek into our hackathon journey.

Here’s to more late-night builds, good vibes, and taking ideas beyond the hackathon.

Till the next all-nighter. 👋