Inspiration

Georgia Tech has thousands of talented students and alumni who have built startups, developed new technologies, and solved difficult problems. But despite having such a strong network, finding the right person to build something with often comes down to luck.

A founder might need an engineer who has experience building medical devices. An alumnus might be looking for a technical cofounder. A student might have exactly the skills a startup needs but never know that opportunity exists.

Georgia Tech already has programs like CREATE-X, Startup Exchange, and alumni networking opportunities. We wanted to build something that complements these resources by making it easier for meaningful connections to happen between Yellow Jackets.

The problem isn't that Georgia Tech lacks talented people. It's that those people don't always know how to find each other.

That's what inspired us to build GT Swarm.

What it does

GT Swarm is a startup matchmaking and collaboration platform that connects Georgia Tech students and alumni based on what they've built, the problems they've solved, and what they want to work on next.

Unlike traditional networking platforms that focus on job titles, majors, or mutual connections, GT Swarm matches people through demonstrated experience.

The platform has several core features:

Find Your People: Founders describe what they need in plain English, and GT Swarm recommends potential cofounders, engineers, advisors, and collaborators based on relevant experience, interests, and availability.

Find Your Startup: Students and alumni receive startup recommendations based on their skills, projects, and the roles they're open to.

GT Swarm - Find Your Startup

Get Unstuck: Users describe a technical or business challenge and discover people who have worked through similar problems before, whether that's Bluetooth firmware, fundraising, or regulatory approval.

Interactive Connection Web: A visual network shows relevant people and the actual relationships connecting them, including potential warm introduction paths through former teammates or collaborators.

GT Swarm - Get Unstuck Connection Web

Explainable Matching: Every recommendation includes a breakdown of why the person or startup is a good fit, instead of relying on an unexplained AI-generated percentage.

GT Swarm - Why This Match

Consent-Based Connections: Users can send invitations, request to join startups, and accept or decline connections before contact information is shared.

One of our biggest priorities was making the platform useful for alumni-to-alumni collaboration, not just traditional student-to-alumni mentorship. We wanted graduates to be able to find other Yellow Jackets who could help them launch companies, solve problems, or pursue new opportunities.

How we built it

We built GT Swarm using Next.js 15, React, TypeScript, and Tailwind CSS, with Supabase PostgreSQL for our hosted database and SQLite for local development.

The most technically interesting part was building our matching and relationship systems.

Our matching engine combines BM25 keyword ranking, a technical synonym system, and optional Google Gemini embeddings to understand how someone's experience relates to a founder's needs. We designed a weighted ranking algorithm that considers relevant work, documented experience, shared Georgia Tech connections, role compatibility, and availability.

We also built a knowledge graph using Graphology, representing relationships among people, projects, problems, and skills. With D3 and SVG, we turned this graph into an interactive connection web. We use Dijkstra's algorithm to identify warm introduction paths through documented relationships rather than assuming two people know each other because they have similar interests.

For onboarding, we implemented document extraction that can turn resumes, project descriptions, and startup materials into draft profiles with the assistance of Google Gemini.

Finally, we deployed the working application on Vercel and developed 72 automated tests using Vitest to check ranking, matching, document extraction, connection workflows, and session behavior.

Challenges we ran into

One of the hardest parts of building GT Swarm was figuring out what actually makes two people worth connecting. Simply attending the same university, having similar interests, or sharing a major doesn't necessarily mean two people would work well together. We wanted our recommendations to reflect what someone could genuinely contribute to a project rather than just how similar their profiles appeared.

To address this, we developed a matching system that considers technical skills, previous projects, problems people have solved, role preferences, and availability. For example, a founder building a medical wearable might need someone with experience in embedded hardware or FDA regulations, rather than simply someone with a biomedical engineering degree.

Another challenge was balancing these factors while making the recommendations understandable. Someone might have the right experience but not be interested in joining a startup. We had to account for these differences and develop our "Why This Match?" feature, which explains the reasoning behind each recommendation.

Ultimately, we learned that meaningful matchmaking isn't just about connecting people with similar backgrounds. It's about finding people whose skills, experiences, and goals complement each other.

Accomplishments that we're proud of

We're proud that we built GT Swarm around a problem we see in our own campus community. Georgia Tech already has incredible students, alumni, researchers, and entrepreneurs, but the right people don't always find each other. We wanted to create something that could bridge that gap and make it easier for Yellow Jackets to connect based on their experiences, skills, and goals.

What excites us most is the potential for GT Swarm to turn those existing connections into real opportunities. Whether it's helping an alumni founder find a cofounder, connecting a student to a startup that needs their skills, or helping someone find an alumnus who has already solved a problem they're facing, we built a platform designed to make collaboration more accessible across Georgia Tech.

We're also proud that we were able to turn this idea into a working product, rather than just presenting a concept. We developed a matching system that goes beyond shared interests to consider what people have actually built and accomplished. Through our interactive connection web, users can see not only who might be a good match, but also how they might be connected through previous projects and collaborations. We wanted every feature to serve a purpose and help users take the next step toward working together.

Ultimately, what we're most proud of is building something that has the potential to continue beyond this hackathon. We see GT Swarm as a platform that could complement the entrepreneurship and alumni programs already at Georgia Tech, bringing together people who might otherwise never meet. Knowing that our project could eventually help fellow Yellow Jackets find teammates, build startups, and bring their ideas to life is what makes this accomplishment meaningful to us.

What we learned

This project taught us that building a useful matching system isn't just about using AI. It's about understanding what makes a recommendation valuable to the person receiving it.

We learned how to combine traditional information retrieval, semantic matching, graph algorithms, and structured data to solve a problem that a simple chatbot or profile directory couldn't fully address.

We also learned the importance of designing around actual user behavior. Finding a good match is only the first step. Users need to understand why that match matters, know how to reach the person, and have a clear way to turn the recommendation into a real conversation.

Most importantly, we learned that sometimes the best technology doesn't create something entirely new. It makes better use of resources and relationships that already exist.

What's next for GT Swarm

Our next step is to launch GT Swarm through a small pilot at Georgia Tech. We plan to reach out to Startup Exchange, CREATE-X, the Student Government Association (SGA), and the Georgia Tech Alumni Association to explore potential partnerships and bring together an initial group of student and alumni founders, entrepreneurs, and professionals.

GT Swarm - Campus Pilot Plan

One of our biggest goals is to strengthen connections within Georgia Tech's alumni network. We want to make it easier for alumni to find potential cofounders, recruit talented students and fellow graduates, share their expertise, and discover startups they can contribute to. At the same time, we want students to have greater access to experienced alumni and meaningful opportunities to work on real projects and startups.

We hope to begin with a small group of active startups and interested alumni and students, using their feedback to improve our matching system and measure the meaningful connections and collaborations that result. From there, we want to expand across Georgia Tech's entrepreneurship programs and alumni community, creating a platform that continues connecting Yellow Jackets long after graduation.

Our long-term vision is to bring this model to other universities, *transforming alumni networks from directories of past graduates into active communities where students and alumni can continue building, collaborating, and creating opportunities together. *

Built With

Share this project:

Updates

Submission history