ClosetBestie: An AI Wardrobe Bestie for My Friend's First Internship

Kritika Mandale — Mon, 05 Oct 2026 00:00:45 +0000

This is a submission for the Hacktoberfest Weekend Challenge: Build for a Friend

What I Built

My friend is about to start her first corporate internship, and there's one little problem that's been making her nervous: "What am I going to wear to the office every day?"

Honestly, girls, I think we can all relate to having a wardrobe full of clothes and still feeling like we have nothing to wear!

She already owns clothes she loves, from sleeveless tops and cute dresses to skirts and frocks. But with a formal office dress code, she wasn't sure how to style them appropriately. Buying an entirely new wardrobe just for an internship didn't feel like the right solution either.

And that gave me an idea.

What if she could create more outfits from the clothes she already owns?

That's how ClosetBestie came to life!

I wanted to build something that feels less like a fashion judge and more like a friend helping you figure out what to wear.

What ClosetBestie Does

Wardrobe Audit: Classifies clothes as Ready, Ready with a fix, or Weekend only based on the user's office rules.
Smart Styling: Suggests practical fixes, like layering a sleeveless top with a blazer when it meets the dress code.
Weekly Outfit Planner: Creates Monday-to-Friday outfit plans with variety and fewer repeated pieces.
Smart Gap Finder: Identifies one versatile item that could unlock more outfit combinations instead of suggesting an entire shopping list.
Gemma 3 Stylist Notes: Adds friendly, encouraging daily notes to make the experience feel more personal.

The goal is simple: more outfit possibilities, fewer unnecessary purchases, and a little more confidence.

Why I Built This

Starting somewhere new is exciting, but it can also be overwhelming. You want to make a good impression, feel comfortable, and find your place. I didn't want choosing an outfit to become another thing my friend had to worry about.

I also didn't want the solution to be, "Just go shopping."

Sometimes, we don't need more clothes. We just need a little help seeing the possibilities in what we already have.

I built ClosetBestie because I wanted my friend to have one less thing to worry about as she starts this new chapter.

It's a small project, but it comes from a real problem. And if it helps someone else feel a little more confident getting ready for their day, that makes it even more meaningful to me.

Try It Out

Live App: https://closetbestie-1.onrender.com/
Backend API Docs: https://closetbestie.onrender.com/docs
GitHub Repository: https://github.com/kritikamandale/ClosetBestie

A Little About the Tech

ClosetBestie combines Next.js, FastAPI, Python, and SQLite with two AI components.

I fine-tuned the open-weight Nemotron model using LoRA and Tinker to understand clothing descriptions. I also use Gemma 3 to generate supportive stylist notes.

In my evaluation on 54 held-out examples, the fine-tuned model improved JSON validity from 13.0% to 51.9% and overall tag accuracy from 9.5% to 47.3% compared with the prompted base model. The results also showed that there's still room for improvement, so I included validation and a manual fallback.

The outfit planner and gap finder use regular Python logic to keep the important rules consistent.

Why Open Innovation Matters

Open-weight models gave me the opportunity to experiment, fine-tune for a specific task, and evaluate the results instead of relying entirely on a general-purpose model.

For me, open innovation is about having the freedom to build around a real problem, learn from what works, and be honest about what still needs improvement.

Prize Categories

Tinker by Thinking Machines: Fine-tuned Nemotron using the Tinker SDK with LoRA and evaluated it against the prompted base model.
Render: Deployed the FastAPI backend and Next.js frontend as separate services using a Render Blueprint.

A Little More Than an Outfit Planner

I know an app can't take away all the nervousness of starting a first internship. But maybe it can make one small part of the experience easier.

Maybe it can help someone rediscover a favourite outfit, try a combination they hadn't considered, or start the morning feeling a little more prepared.

That's what I wanted to build for my friend. Not another reason to buy something new, but a little reminder that what you already own can be enough.

More outfits from the clothes you already own. Fewer unnecessary purchases. And a little more confidence for whatever comes next.

DEV Community: Kritika Mandale