Inspiration
This was a fixed requirement from the developed solution... so not much for inspiration front up!
What it does
Plutus analyzes an HVAC contractor’s project portfolio to identify financially risky projects, explain why they are going wrong, and propose concrete recovery actions.
How we built it
We split the system into five stages: loading and cleaning the data, turning it into a structured working dataset, flagging questionable projects, running LLM analysis through three separate agents in three separate stages, and then packaging the results behind a frontend built for quick executive review.
Challenges we ran into
Git coordination was a major challenge. Not everyone on the team was equally comfortable with branching and merging, so we ran into constant conflicts, syncing issues, and delays waiting for each other’s changes
Vercel was also difficult to set up because none of us had much prior experience with its ecosystem, and we struggled with team access and ownership while trying to use v0 effectively.
Combining everyone’s parallel work at the end was messy and much harder than expected, and we ended up relying heavily on AI to untangle and integrate our (drastically) different workflows together,
Accomplishments that we're proud of
We’re proud that we managed to use the Vercel and v0 tooling meaningfully instead of treating it like a formality!
What we learned
Orchestrating a coding group for a project is like orchestrating a cluster of GPUs, you want more parallel processes for higher throughput for each group member!
What's next for Plutus - Return Your Wealth
If we actually were awarded some rankings in this hackathon... would we have some motivation to keep developing this little project?
Built With
- duckdb
- fastapi
- next.js
- pandas
Log in or sign up for Devpost to join the conversation.