IntelliAdmit - AI-Powered Admission Automation

A multimodal AI agent transforming student inquiries into verified leads with automated cloud synchronization.

Comment

InspirationThe inspiration for IntelliAdmit came from observing the "Information Gap" in college admissions. At SNS College of Technology, students often have urgent questions at night or during weekends when staff are unavailable. I wanted to build an AI that acts as a 24/7 professional consultant—not just answering questions, but actively capturing data and delivering official documents instantly.

What it does

IntelliAdmit is a multimodal AI agent that automates the first layer of student inquiries. It answers questions about engineering streams like CSE, IT, and AIML while providing interleaved visual imagery of the campus. Most importantly, it captures student details (Name, Email, Phone, Domain) and synchronizes them to an Aiven MySQL cloud database, while simultaneously emailing a professional fee structure to the student.

How we built it

The system is built on a cloud-native architecture. We used Node.js and WebSockets for real-time communication, hosted on Google Cloud Run. The AI logic is powered by the Gemini 1.5 Flash model via the Google GenAI SDK. For data persistence, we utilized Aiven MySQL, and for the automated communication layer, we integrated Nodemailer to trigger branded HTML emails.

Challenges we ran into

We encountered several engineering hurdles. First, we faced an ENOTFOUND error where the cloud server couldn't resolve the database hostname; we solved this by optimizing the connection pool with IPv4 family prioritization, or$$family: 4$$. Second, we hit API rate limits ($$429$$errors), which we bypassed by engineering a fail-safe fallback system that ensures lead capture even if the AI model is temporarily throttled.

Accomplishments that we're proud of

We are proud of achieving Zero Data Loss. By implementing a robust backend validation layer, we ensured that every single student inquiry is saved to the database regardless of traffic spikes. We are also proud of the high level of professionalism in the automated email responses, which feature a dynamic fee table that matches the college's official branding.

What we learned

This project taught us the critical importance of System Resilience. We learned that building a successful AI tool requires more than just a good prompt; it requires a stable infrastructure. We mastered the nuances of Cloud Run deployments, managed database handshakes via SSL, and the complexities of Asynchronous tool-calling within the Gemini ecosystem.

What's next for IntelliAdmit - AI-Powered Admission Automation

The next phase for IntelliAdmit involves integrating OCR technology for automated marksheet verification and porting the entire logic to WhatsApp Business API for better accessibility. We also plan to build an Admin Dashboard in React to allow admission officers to visualize and export the student data directly from our MySQL cloud storage.

Built With

Share this project:

Updates

Submission history