Inspiration

Every student I know spends hours tailoring resumes for each job application. They copy paste, reword bullets, try to guess what keywords the ATS wants. It's exhausting and most of the time the resume still doesn't get through screening. I wanted to build something that does the heavy lifting for you. Paste the job description, let AI read your full background, and get a resume that is actually written for that specific role. No subscriptions, no accounts, no company storing your career history on their servers.

What it does

Rohero is a locally running AI resume builder that tailors your resume to every job you apply for. You fill in your profile once with all your experience, projects, skills and education. Then for each application you paste the job description and Rohero does three things. First it analyzes the gap between your current skills and what the job needs, showing you exactly what is missing and suggesting projects you could mention. Second it calls the Gemini AI to read your full profile and the job description together, selects the most relevant experience and projects, rewrites your bullets with strong action verbs and keywords from the JD, and generates a complete ATS optimized resume. Third it exports a clean PDF ready to submit. Every application you send gets tracked in the built in job tracker. When you have interviews lined up the interview prep section pulls company information, recent news, and generates technical and behavioral questions specific to that role and company. All of this runs in your browser. No login. No cloud storage. Your data never leaves your machine.

How we built it

Rohero is built with React and Vite on the frontend with Tailwind CSS for styling. The Gemini API powers all AI features including resume parsing, skills gap analysis, resume generation, and interview prep. We use jsPDF and html2canvas to convert the live HTML resume preview into a downloadable PDF. Job application data and the user profile are stored in localStorage and IndexedDB so everything persists locally without any backend. The app has a built in API key rotation system that falls back from Gemini to Groq to NVIDIA NIM if any provider hits a rate limit, so the experience never breaks during a demo or heavy usage.

Challenges we ran into

The biggest challenge was getting the Gemini API to return clean structured JSON consistently enough to render a real resume from. LLMs occasionally return malformed JSON or wrap it in markdown code blocks, so we built a robust parsing layer that strips fences, handles partial responses, and retries on failure. Getting the PDF export to look exactly like the HTML preview was also tricky since html2canvas renders fonts and spacing slightly differently at different scales. The API rate limiting on the free tier was a real blocker during development and pushed us to build the multi provider fallback system which ended up being one of the strongest features of the app.

Accomplishments that we're proud of

We are proud that the core loop actually works end to end. A user can fill in their profile, paste any real job description, and get back a genuine tailored resume with relevant bullets and keywords in under 15 seconds. The skills gap analysis gives actionable feedback that most resume tools charge money for. Building a fully local app with zero backend that still feels fast and polished is something we are genuinely proud of. And the multi provider AI fallback means the app keeps working even when one API hits its limit.

What we learned

Prompt engineering is harder than it looks. Getting an LLM to return perfectly structured JSON every single time required more iteration than any other part of the build. We also learned that local first architecture is underrated. Removing the backend completely simplified the whole project and made the privacy story genuinely strong rather than just a marketing claim. Working under a 24 hour deadline taught us to cut scope ruthlessly and ship the core loop first before adding anything else.

What's next for Rohero

The next version of Rohero will ship as an Electron desktop app so the PDF resumes save directly to a local folder rather than the downloads directory. We plan to add field specific templates for non CS students including life sciences, civil engineering, business and creative fields since most resume tools are built only for software engineers. The interview prep section will get richer with company specific question banks. And we want to add a resume version comparison feature so users can see which resume version gets the most callbacks over time.

Built With

Share this project:

Updates