Inspiration
As a student in Bangladesh, I saw classmates graduate with strong degrees but struggle to land tech jobs. Traditional resumes only show claims, not proof. Universities do not know what modern industry needs, and recruiters spend weeks filtering unverified applicants. I built CareerBridge to create a transparent, verified bridge connecting students, mentors, universities, and employers.
What it does
CareerBridge connects four key stakeholders in one collaborative web ecosystem:
- Students: Upload resumes (PDF, image, or text) to extract complete skills and projects without truncation, visualize gaps on an interactive 5-axis radar chart, take AI Skill Audits, follow tailored Bridge Projects, and apply to matched jobs.
- Mentors: Review mentee requests, guide learning through 4 structured assessment types (video lectures, practicals, quizzes, assignments), grade tasks in sessions, earn XP, and unlock an official 500 XP certificate.
- Universities: Automatically view cohort skill gaps linked via institutional emails and receive AI-generated curriculum improvement reports.
- Industry Recruiters: Post jobs, create custom screening forms, and review candidates pre-ranked by an algorithmic match score.
Transparent Job-Skill Match Formula
Instead of arbitrary scores, CareerBridge calculates match percentages using a transparent model with LaTeX: $$\text{Match}(\%) = \text{clamp}(B + E + (0.4 \times A), 0, 100)$$
- B (Baseline): Accounts for degree rigor and academic background.
- E (Direct Matches): Rewards verified skills matching the job.
- 0.4 x A (Synergy): Rewards transferable knowledge (e.g., core programming helping you learn backend faster).
- clamp: Binds the final score realistically between 0% and 100%.
What's new for this buildathon
| Area | Before Buildathon | Now (After Buildathon) |
|---|---|---|
| AI Architecture | Client-side API calls exposed in the browser. | Full server-side proxy (/api/gemini/*) with zero exposed keys and automatic fallback responses. |
| Role Taxonomy | Static, hardcoded skill lists. | Dynamic AI taxonomy generating 5-axis radar benchmarks for tech and non-tech roles. |
| Skill Verification | Self-reported resume claims. | Interactive AI Skill Audits and tailored Bridge Projects with estimated hours to close gaps. |
| Mentorship | Basic static mentor list. | Full lifecycle workspace: milestone tasks, grading, 1-on-1 chat, XP Leaderboard, and 500 XP certificates. |
| Hiring Pipeline | Simple job board view. | Custom form builder, 1-click apply, and live status tracking (Under Review, Interview, Offer). |
How we built it
- Frontend: React 19, TypeScript, Tailwind CSS, Recharts.
- Backend: Node.js and Express running server-side with Vite middleware.
- Database & Auth: Firebase Firestore for real-time synchronization and Firebase Authentication.
- AI Integration: Google Gemini API (
@google/genaiusinggemini-3.8-flash) orchestrated securely on the server with deterministic fallbacks.
Challenges we ran into
- Deep Multimodal CV Analysis Latency: Because the AI deeply analyzes complex, unstructured PDF/image resumes to extract every project, skill, and gap without losing any detail, the analysis API call takes around 1 to 2 minutes to process completely.
(Note for judges and evaluators testing the live site: When uploading a CV for analysis, please patiently allow 1-2 minutes for the comprehensive evaluation to finish). - API Resilience: Server endpoints threw permission errors when API keys were missing in preview environments. I built an intelligent fallback system that serves deterministic role taxonomies and benchmarks so the app never crashes.
- Solo Architecture: Managing real-time data sync across four distinct user roles in Firestore required strict state management and security rules.
Accomplishments that we're proud of
- Built a functional 4-sided platform entirely as a solo developer.
- Created a transparent mathematical scoring model ($B + E + 0.4A$) instead of arbitrary black-box AI scores.
What we learned
I learned how to migrate client-side AI calls into a production-grade Express backend and manage complex multi-role state machines in Firestore.
What's next for CareerBridge
- Adding verifiable, cryptographically signed certificates for completed mentorships.
- Partnering with local university departments in Bangladesh for a live pilot cohort.
Try it out
- Live App: https://careerbridge-2026.netlify.app
- Demo Accounts: Student email- [email protected] ; Password- abcd1234 University email- [email protected] ; Password- abcd*1234 Mentor email- [email protected] ; password- abcd1234 Industry email- [email protected] Password- abcd*1234
- GitHub Repo: https://github.com/moni408/CareerBridge2026.git
- Demo Video: https://youtu.be/Yw0U8hAsKX4?si=UYzoQ_FYin8brCXI
Built With
- express.js
- firebase
- gemini-api
- lucidreact
- node.js
- react
- tailwindcss
- typescript