Jimmy

Landing Page
About Jimmy
Jimmy in Mocking Mode
Jimmy's Feedback

Inspiration

We’ve all been there—staring at a Zoom screen, sweaty palms, giving robotic answers to a recruiter who sounds like they’ve asked the same questions a thousand times. That’s when it hit us: why not simulate that pressure, but with an AI recruiter that’s trained not just to ask, but to evaluate, guide, and help you improve? Enter Jimmy—our virtual interviewer who’s sharp, structured, and always on the clock (literally). Jimmy doesn't just mock interview you—Jimmy trains you.

What it does

Jimmy is your AI-powered interview coach. Upload your resume, paste the job description, and start a live mock interview with Jimmy. He reads your documents, asks recruiter-style questions using real company needs, and dynamically adjusts based on your responses. After the interview, Jimmy evaluates your performance across multiple criteria—clarity, relevance, behavioral depth, and STAR structure—and gives you precise feedback, so you're always one step closer to landing the job.

How we built it

We built Jimmy with a stack that balances performance, flexibility, and interactivity. The backend is powered by FastAPI, using Cohere’s RAG pipeline for smart context-aware generation. We trained Jimmy on real recruiter transcripts and fine-tuned prompt engineering so that it behaves less like ChatGPT and more like a human interviewer. The frontend was built using React, TypeScript, and TailwindCSS, with Web Speech API to allow voice-based interactions for an immersive, human-like experience.

Challenges we ran into

Teaching Jimmy to behave like a real recruiter was tough. The language model is powerful, but by default, it’s more ChatGPT than corporate HR. We spent hours iterating on prompt engineering, testing various scenarios to make Jimmy avoid being too casual, overly agreeable, or redundant. Getting it to ask "Tell me about yourself" at the right time? Harder than it sounds. Aligning the retrieval logic and prompt injection to guide its tone and timing was a real mental workout.

Accomplishments that we're proud of

We shipped a fully functioning MVP with voice interaction, resume/job description parsing, RAG-driven context awareness, and detailed interview feedback—all in under 36 hours. Jimmy doesn’t just talk to users—he understands their documents and reflects that in real-time conversation. From a simple upload to intelligent back-and-forth, we made something actually useful—and a little charming too.

What we learned

We learned how to use Cohere’s embeddings effectively for retrieval, how to handle voice input/output in the browser, and most importantly—how to craft AI experiences that feel real. Building Jimmy made us better developers, better problem solvers, and slightly better interviewees.

What's next for Jimmy

Jimmy that would crack you on a real technical interview with some coding involved.
Company-specific simulation modes (e.g. Google, Goldman Sachs recruiter).
Female Jimmy? Jimmy in a different outfit? Moving Jimmy with voice imitation? Stay tuned.
And of course, Jimmy with jokes—only when appropriate.