CareerCraft AI

CareerCraftAI is an AI-powered career assistance platform designed to help job seekers and professionals streamline their job search and career growth.

Created on 8th April 2025

•

Manual Resume Building was Time-Consuming & Inconsistent
🔧 Solution: Built an AI-powered resume and cover letter generator that customizes content based on job titles and skills — saving time and improving personalization.
Resumes Getting Rejected by Applicant Tracking Systems (ATS)
🔧 Solution: Integrated ATS optimization logic, ensuring resumes are formatted for compatibility and keyword-rich based on job descriptions.
Job Seekers Lacked Real-Time Market Trends
🔧 Solution: Incorporated job market insights using AI to provide analytics on in-demand roles, skills, and industry trends — helping users make smarter career decisions.
Candidates Felt Unprepared for Interviews
🔧 Solution: Added AI-generated interview questions and tips, simulating interview prep tailored to roles — boosting confidence and performance.
User Authentication was Complex to Implement Securely
🔧 Solution: Integrated Clerk for seamless, secure authentication and user management — ensuring data privacy without friction.
Disjointed UI/UX in Career Platforms
🔧 Solution: Built a modern and intuitive frontend using Next.js + shadcn/ui, enhancing user experience and accessibility across devices.

🔐 Clerk Authentication Integration Conflicts
When integrating Clerk with Next.js and custom API routes, I faced issues with session persistence and accessing user metadata across server and client.
✅ Fix: I carefully aligned Clerk’s
withServerAuth
and
useUser()
hooks and refactored my API routes to securely handle auth tokens. Also, enabling SSR support properly resolved hydration mismatches.
📄 Resume Formatting for ATS Compatibility
While generating resumes using AI, the output was rich in content but not always ATS-friendly (e.g., tables, headers, or font choices caused parsing issues).
✅ Fix: I studied ATS parsing standards and implemented a formatting filter layer to strip unnecessary styles and generate clean, semantic HTML and PDF outputs.
⚙️ Prisma + Supabase Sync Issues
During initial development, I faced schema sync problems where changes in Prisma models weren’t reflected correctly in Supabase.
✅ Fix: I introduced migration versioning and rollback scripts, and established a workflow using
prisma migrate
followed by schema validation checks on deployment.
🧠 Gemini AI Prompt Tuning
The AI initially generated vague resume content or job recommendations.
✅ Fix: I tuned the prompts using role-specific templates and added contextual tags (skills, job level, industry) to get more accurate and personalized outputs.

Technologies used

Node.js

Next.js

Express.js

Tailwind CSS

React.js

Supabase

Clerk

Prisma ORM

shadcn/ui

Google Gemini AI