Inspiration
Freelancers and clients often face challenges with milestone-based payments. After completing a task, freelancers may have to wait while clients manually review their work, verify requirements, and approve payments. This process can be time-consuming, inconsistent, and sometimes lead to disputes.
We wanted to explore a simple question: What if AI could verify completed technical work and connect that verification directly to the payment process?
This inspired us to build NexusPay AI — From Pull Request to Payday, an AI-powered platform that connects GitHub code verification with PayPal-based milestone payments.
What it does
NexusPay AI is designed to automate technical milestone verification and simplify freelance payment decisions.
The workflow is straightforward:
Create a Milestone: The client enters the project requirements, milestone description, payment amount, freelancer's PayPal email, and GitHub pull request URL.
AI Code Verification: NexusPay AI retrieves the GitHub code changes and uses Groq-powered AI to compare the implementation against the specified requirements.
Confidence Score and Report: The AI generates a confidence score, explains its findings, and recommends approval or rejection.
Automated Payment Workflow: Approved milestones can initiate a PayPal Sandbox payout request.
Client Override: If the AI rejects a milestone, the client can review the findings and provide a reason to override the decision.
Dashboard: The application displays milestone activity, verification results, and payment statuses.
The goal is to reduce manual code-review effort while keeping clients informed and in control of payment decisions.
How we built it
We developed NexusPay AI as a full-stack web application, connecting several technologies through a unified workflow.
Technologies used:
React and Tailwind CSS: For the frontend interface and dashboard.
Node.js backend: To coordinate milestone processing and external API integrations.
GitHub API: To retrieve pull request changes for verification.
Groq API: To analyze code using AI and generate structured verification reports.
PayPal Payouts API and Sandbox: To integrate and test milestone-based payment requests without real-money transactions.
Vercel: To deploy the web application.
The backend connects GitHub code retrieval, AI evaluation, milestone decision-making, and payment processing into one workflow.
Challenges we ran into
One of our biggest challenges was ensuring consistent and reliable AI verification.
During testing, the same GitHub pull request sometimes received different verification results. This highlighted the limitations of using probabilistic AI assessments for decisions involving payments.
Another significant challenge was integrating PayPal Sandbox payouts. While we implemented the payout workflow, some test payout requests failed, requiring further investigation into API responses, sandbox configuration, and error handling.
We also explored how to balance automation with human oversight. This motivated the Client Override feature, allowing clients to make the final decision rather than relying entirely on AI.
Accomplishments that we're proud of
Built and deployed a functional web application.
Integrated GitHub pull request analysis with Groq-powered AI verification.
Implemented confidence scoring and detailed technical evaluation reports.
Developed both approval and rejection workflows.
Added a reason-based client override feature.
Integrated a PayPal Sandbox payout workflow for testing.
Created a dashboard for monitoring milestones and their statuses.
Our biggest accomplishment was bringing code review, AI-assisted decisions, and payment authorization together within a single application.
What we learned
Building NexusPay AI helped us understand how to combine AI-powered analysis with real-world financial workflows.
We learned about GitHub API integration, structured AI responses, asynchronous backend operations, payment APIs, and debugging failures across multiple services.
Most importantly, we learned that AI should assist financial decisions, not make irreversible decisions without appropriate safeguards.
Reliable verification, human oversight, transparent explanations, and careful payment handling are essential when building systems involving money.
What's next for NexusPay AI — AI-Powered Freelance Payments
We plan to improve NexusPay AI through:
More consistent and reliable AI verification.
Successful end-to-end PayPal Sandbox payout validation.
Stronger authentication and payment authorization controls.
Automated test execution alongside AI code review.
Support for multiple milestones and freelancer projects.
Better error handling, payment tracking, and transaction history.
Freelancer and client dashboards with notifications.
Improved safeguards against duplicate payouts.
Our long-term vision is to create a trustworthy platform where verified technical work can move efficiently from submission to payment.
NexusPay AI — From Pull Request to Payday.
Built With
- express.js
- githubapi
- groqapi
- javascript
- mongodb
- mongoose
- node.js
- paypalpayoutsapi
- paypalsandbox
- react
- restapi
- vercel
- vite
