CampusSphere

AI-powered ecosystem that aggregates developer opportunities and uses LPU inference to generate custom project roadmaps and resume match scores.

Comment

Inspiration

In 2026, students aren't suffering from a lack of opportunity; they are suffering from Information Fragmentation. With 50+ platforms hosting contests, internships, and hackathons, the "mental tax" of tracking deadlines leads to burnout.

We built CampusSphere AI to move the needle from "finding an event" to "winning an event." We wanted a command center that doesn't just aggregate data but generates actionable intelligence.

What it does

CampusSphere AI acts as a centralized, intelligent hub for Hackathons, Internships, and Coding Contests. It solves the fragmentation problem by replacing static lists with dynamic strategy:

  • The Blueprint Engine (Hero Feature):
    It analyzes event requirements against your specific skill set to generate a day-by-day, step-by-step technical roadmap to win.

  • Smart Matcher:
    A Groq-powered parsing engine that assigns a Win-Probability Score by analyzing your resume's technical depth against active opportunities.

  • T-Minus Protocol:
    A proactive defense system that triggers automated, multi-channel alerts (Email/Push) exactly 120 minutes before a contest launch.

  • Global Discovery:
    A high-speed interface utilizing a 500ms debounce mechanism to find opportunities based on semantic context, not just rigid keywords.

How we built it

We architected the system using a Service-Oriented Architecture (SOA) to ensure maximum speed and modularity:

  • Frontend: Built with React.js and Tailwind CSS, featuring a Glassmorphic "Bento Box" UI to maximize information density.
  • AI Inference: Powered by Groq Cloud (Llama 3) to achieve near-instant (800+ tokens/sec) strategy generation.
  • Database: MongoDB Atlas with advanced aggregation pipelines for real-time event filtering.

The Match Algorithm

We calculate the alignment score ( S ) using:

[ S = \left( \frac{|R \cap E|}{|E|} \right) \cdot \omega + \delta(T) ]

Where:

  • ( R ) = Extracted Resume Skills
  • ( E ) = Event Requirements
  • ( \omega ) = Technical Priority Weight
  • ( \delta(T) ) = Time-decay factor for approaching deadlines

Challenges we ran into

The biggest hurdle was Data Sanitization. Extracting text from stylized, multi-column PDF resumes using pdf-parse often resulted in unusable "word salad."

We solved this by building a Pre-Processing Pipeline that passes raw text through Groq to rectify data into structured JSON before it hits our engine:

const rectifiedData = await groq.chat.completions.create({
  messages: [
    {
      role: "system",
      content: "Convert this unstructured PDF text into a valid skill-set JSON schema."
    }
  ],
  model: "llama3-70b-8192",
});

We also tackled Regex limitations. Literal string matching missed relevant events (e.g., searching "Frontend" but missing "React"). We pivoted to a pseudo-semantic approach, mapping categories to tech stacks to ensure higher discovery rates.

Accomplishments that we're proud of

  • Functional Strategy Engine:
    We moved the Roadmap feature from a concept to a fully functional tool that provides specific, non-generic advice for every event.

  • Extreme Performance:
    Reducing the generation time for a 5-day hackathon strategy from 15 seconds to under 1.8 seconds.

  • Zero-Drift Automation:
    The T-Minus Protocol successfully handles global time-zone conversions across LeetCode, Codeforces, and Devpost without manual intervention.

📖 What we learned

We learned that in 2026, Speed is a UX feature. Users won't wait for a slow AI spinner; the "magic" only happens when the response is instantaneous.

We also realized that providing a student with a link is a utility, but providing them with a strategy is a partnership.

What's next for CampusSphere AI

  • Peer Matchmaking:
    Using our AI logic to suggest teammates whose skills complement your specific gaps for a hackathon.

  • GitHub Integration:
    Directly pulling repository data to verify skill claims and improve match accuracy.

  • Browser Extension:
    To Smart-Match opportunities in real-time as users browse LinkedIn or other job portals.

Share this project:

Updates