Inspiration
I saw many CS freshers around me struggle to get their first job. Every job site asks for experience, and to get ready you need many different tools: one site for jobs, one for resume checks, one for interview practice. It is tiring, and many people give up. I wanted to build one free place that does all of it.
What it does
TalentLaunch helps CS freshers get their first job.
- Job search: collects openings from about 20 sources at once, removes duplicates and marks roles that suit freshers.
- 28 AI tools: resume check, cover letter, resume "roast", offer comparison, salary help and more.
- Voice interview: the AI asks questions out loud and you answer by speaking, then you get feedback.
- Tracker: save jobs, track each application and export the list as a PDF.
- Motivation: points, badges and daily streaks keep you going.
- Extras: dark, light and neon themes, 4 languages (English, Hindi, Tamil, Telugu) and an installable web app (PWA).
How I built it
- Frontend and backend: Next.js 14 (App Router), TypeScript and Tailwind CSS.
- Database and login: MongoDB Atlas with Mongoose, JWT login, and roles for student, recruiter and admin.
- Jobs: I call many job APIs in parallel with
Promise.allSettled, set a time limit for slow ones, and remove duplicates by title and company. - AI: five free AI providers (Groq, Gemini, Cohere, Mistral, OpenAI) behind one adapter. If one fails, the next one answers automatically.
- Voice: the browser's Web Speech API for listening and speaking.
- Share cards: drawn on the Canvas API as images.
Challenges I ran into
- Slow or broken job APIs. Some sources time out or fail. I fixed this by never letting one source block the others and by keeping a backup list of jobs.
- Free AI limits. Free plans run out or fail often. The fallback chain across five providers solved it.
- Messy AI output. The AI sometimes returned broken JSON. I ask for a strict format and clean the answer before reading it.
- Setup problems. Environment variables, CSP rules in the browser extension and reading PDF files all took time to debug.
Accomplishments that I'm proud of
- A working app that is free to run, with many features in one place.
- An AI system that keeps working even when a provider goes down.
What I learned
- Building for failure: using
Promise.allSettled, time limits and fallbacks. - Designing roles and secure login, and working with many APIs.
- Making a product simple enough that a fresher can use it on day one.
What's next for TalentLaunch
- Move login tokens to secure cookies and add rate limiting.
- Save jobs in the database instead of memory.
- Email alerts for new fresher jobs and for campus drives.
- A recruiter chat and a placement-cell portal for colleges.
Built With
- bcrypt
- canvas-api
- cohere
- google-gemini
- groq
- javascript
- jwt
- mistral
- mongodb
- mongodb-atlas
- mongoose
- netlify
- nextjs
- node.js
- openai
- pwa
- react
- rest-api
- tailwindcss
- typescript
- vercel
- web-speech-api