Backer

Inspiration

The reality is that great ideas are everywhere, but venture capital is usually stuck in a few specific zip codes. As CS students, we keep seeing the same thing: founders in underserved regions aren't failing because they lack talent; they’re failing because they lack a network.

In these "geographical blind spots," you can have the best product in the world, but it dies in an inbox because cold emails don't work. We built Backer to kill that "postal code ceiling." We wanted a platform where your grit and your code matter more than where you're logging in from.

What it does

Backer is basically a global bridge that cuts out the networking middleman.

For Founders: It’s a stage. Instead of writing formal pitch decks that nobody reads, you post high-impact "product reels" that show off your actual work and experience.

For Investors: It’s a direct window into global talent. They scroll a personalized feed of startups, pledge support, and jump straight into a chat.

We stripped away the administrative hurdles so founders can stop chasing signatures and actually focus on building their product.

How we built it

We wanted this to feel like a production-ready tool, so we went with an AWS-native architecture designed for real-time discovery.

The core is our Intelligent Feed Ranking System. We didn't want a static list; we wanted a feed that actually learns. We built a retrieval-and-ranking pipeline on top of DynamoDB that scores projects using a few key signals:

Behavioral Affinity: What the investor has liked or committed to before.

Semantic Relevance: We use NLP tokens to match the startup’s "problem/solution" description with the investor's interests.

Freshness & Exploration: We use a decay function to keep the feed from getting stale while forcing the algorithm to "explore" new founders so they get a fair shot.

Basically, it's a foundation that’s ready for a full ML model, but currently uses real-time behavioral signals to keep the feed high-intent and relevant.

Challenges we ran into

Delegating the work to each team member was difficult since a lot of the project depended on others. This slowed down our production overall. In addition to this, as we were developing parts of the program separately, we found difficulties in the integration of each part. We also went through many iterations of the design process, trying to make sure we understood the end goal of the product and aiming to stick to the project's main goals.

Accomplishments that we're proud of

As a collective, we were able to build a fully functional end-to-end system that has the minimum functionality we aimed for at the beginning. It was very rewarding to also see the scrolling feature of the short-form videos getting to work, as this is a main component of most of the popular social media apps today.

What we learned

For some members, this was their first time using AWS services and building a full-stack application. We also learned about the planning stages of building a full system, thinking about various design decisions.