Inspiration
As a Computer Science student, I constantly saw my peers stuck in "Tutorial Hell." We all want to become Data Scientists, AI Engineers, or Full Stack Developers, but we are paralyzed by choice. There are 10,000 courses on the internet. Some cost ₹500, some cost ₹50,000. Some are great, some are outdated. Most students end up buying expensive boot camps just to get a syllabus, or they blindly follow random YouTube videos without a structured path. I asked myself: Why isn't there a tool that acts like a senior mentor? One that knows my specific budget, my time constraints, and builds a custom curriculum for me instantly? That frustration led to Eduvo AI—the first AI Career Architect designed to democratize high-quality mentorship.
What it does
Eduvo AI is a full-stack, intelligent career guidance platform. It doesn't just "chat" with you; it builds a workspace for your career. Hyper-Personalized Roadmaps: Users input their career goal (e.g., "AI Architect"), current experience, weekly availability, and exact budget (including "₹0 / Free Only"). Interactive Timeline: The AI generates a structured, step-by-step roadmap with specific topics to learn week-by-week. Smart Course Finder: My proprietary "Affiliate Injection Engine" scans the roadmap and automatically recommends the best-rated courses (Udemy, Coursera, etc.) that match the user's specific budget. Gamified Dashboard: To solve the dropout rate, I built a dashboard that tracks daily streaks, progress percentages, and unlocks "Badges" (like 'First Step Taken') to keep students motivated. Persistence: Users can save roadmaps to their profile, track progress over months, and even download their plan as a PDF.
How I built it
I architected Eduvo AI as a Serverless, High-Performance Single Page Application (SPA) to keep costs at zero while scaling infinitely. The Brain (AI): I used Google Gemini Pro API. The biggest engineering challenge was prompt engineering. I designed a complex "System Instruction" using few-shot prompting to force the LLM to output strict, validated JSON data instead of unstructured text. This allows the frontend to render interactive UI elements rather than just a text block. The Backend: Firebase. I used Firebase Authentication (Google OAuth) for secure logins and Cloud Firestore (NoSQL) to store user profiles, roadmaps, and gamification stats in real-time. The Frontend: Built with Vanilla JavaScript and Tailwind CSS. I avoided heavy frameworks like React to ensure a 100/100 Google Lighthouse performance score and instant load times. The Interface: I implemented a "Glassmorphism" UI with a 3D floating Aurora background using CSS animations to give it a futuristic, premium feel.
Challenges I ran into
AI Hallucinations & JSON Parsing: Initially, Gemini would return markdown or conversational text, which broke the roadmap renderer. I had to implement robust error handling and strict JSON schema enforcement in the prompt to guarantee the app never crashes. Monetization Logic: I wanted the app to be free for students but sustainable as a business. Building the logic to programmatically match a generic AI step (e.g., "Learn Python") with a specific revenue-generating affiliate link without ruining the UX was tricky. DevOps & Security: Configuring Firestore Security Rules (RBAC) to ensure users could only delete or read their own data took significant testing. I also had to learn DNS management (SPF/DKIM records) to ensure transactional emails didn't land in spam.
Accomplishments that I'm proud of
It's Live: I didn't just build a localhost toy. I deployed it to a custom domain (www.eduvoai.com) with full SSL and CDN distribution via Netlify. The "Budget" Algorithm: I’m proud of the logic that respects the user's currency (INR/USD). If a user selects "Low Budget," the AI genuinely prioritizes free documentation and YouTube playlists over paid courses. Gamification Engine: I built the streak counter and badge system from scratch using local storage and database timestamps, which has arguably increased user retention in my initial testing. Performance: Achieving a near-perfect 100/100 Performance Score on Google Lighthouse by optimizing assets and code splitting.
What I learned
This project transitioned me from a "Coder" to a "Product Engineer." I learned that Prompt Engineering is less about "asking nicely" and more about defining strict data structures. I mastered Firebase security rules and database modeling. I learned the business side of tech: SEO, Affiliate Marketing, and Domain Management are just as important as the code itself. Most importantly, I learned that the best products solve your own problems.
What's next for EduvoAI
Resume Builder: An AI tool that takes the user's completed roadmap and auto-generates a resume highlighting those specific skills. AI Mock Interviews: Integrating a Voice API to let users practice technical interviews for the specific role they generated a roadmap for. Community: Adding a "Leaderboard" feature to let students compete on learning streaks.
Log in or sign up for Devpost to join the conversation.