-
-
Resume Explorer to help generate roles
-
Career explorer to display related job titles
-
interview prep view with the fields
-
Resume uploaded and saved for the user session
-
Job Analyzer with fields populated before analysis
-
Career explorer with related job titles based on the resume
-
Job Analyzer with match score of the job
-
Resume Studio generated Cover letter
-
Resume Studio with pionts to improve resume
-
Learning path with generated courses
-
Career Path Explorer with the related job titles
-
Archtectural Diagram
Inspiration
Job seeking is an incredibly stressful and often inefficient process. Many talented individuals struggle to articulate their skills effectively on a resume, tailor their applications to specific roles, or perform well in high-stakes interviews. We were inspired to build a single, intelligent tool that acts as a personal career coach, leveraging the power of Gemini to demystify the job application lifecycle and empower users to put their best foot forward. We wanted to create an "AI copilot" that assists with every step, from exploring new career paths to negotiating a final salary offer.
What it does
AI Career Copilot is a comprehensive, AI-powered web application designed to assist job seekers. Its key features include:
My Resume: A central place to upload and store a user's resume for use across all modules. Career Explorer: Analyzes a resume to suggest alternative career paths and related jobs the user might be qualified for.
Job Analyzer: Scores a resume against a specific job description, provides actionable feedback for improvement, and generates a tailored cover letter. Resume Studio: Transforms raw, unstructured notes about a user's experience into polished, professional, STAR-method resume bullet points.
Learning Pathfinder: Identifies skill gaps between a user's resume and a target job role, then generates a personalized learning plan with links to relevant resources.
Interview Prep: A mock interview simulator with two modes: a text-based interview for traditional practice and a real-time, voice-based conversational interview for a more realistic experience. It provides detailed performance feedback after the session.
Salary Strategist: Researches market data for a specific role and location to generate a data-driven salary negotiation strategy.
How we built it
The application is built with a modern, serverless architecture, aligning with the hackathon's focus on Google Cloud Run.
Frontend: A responsive single-page application built with React and TypeScript, styled with Tailwind CSS for a clean, modern user interface. It runs directly in the browser and is hosted on Firebase Hosting. Backend: The application runs entirely on the client-side, making direct calls to the Gemini API. This simplifies the architecture and removes the need for a separate backend service, demonstrating the power of modern web technologies and direct AI integration. AI Integration: The core logic is powered by the Google Gemini API. We use a combination of models: gemini-2.5-flash for fast, efficient tasks like generating resume bullets and interview questions, and gemini-2.5-pro for more complex, nuanced analyses like job matching and interview feedback. The real-time voice interview feature uses the Live API (gemini-2.5-flash-native-audio-preview-09-2025) for low-latency, conversational interaction. Infrastructure: The entire application is hosted on Firebase Hosting, providing a global CDN for fast delivery of the static assets (HTML, CSS, JS).
Challenges we ran into
One of the most significant challenges was implementing the real-time voice interview. Managing audio streams from the microphone, encoding it into the required PCM format, sending it to the Gemini Live API, and then receiving, decoding, and playing back the AI's audio response in a seamless, low-latency stream required careful handling of Web Audio APIs. We had to ensure perfect synchronization to avoid choppy audio or delays, and we also implemented robust transcription handling to display the conversation in real-time. Another challenge was designing effective prompts and JSON schemas for the Gemini models to ensure consistent, reliable, and structured data for every feature. It took several iterations to get the models to return data in the exact format we needed for the UI, especially for complex outputs like the interview performance report.
Accomplishments that we're proud of
We are incredibly proud of the voice-based interview prep feature. Creating a fluid, real-time conversational experience with an AI is a complex task, and seeing it come to life was a major accomplishment. It feels like a truly next-generation tool for interview practice. We're also proud of the application's comprehensive nature. We didn't just build one tool; we built an entire suite of integrated services that cover the job application process from start to finish. The way the user's resume seamlessly integrates into each feature creates a cohesive and powerful user experience.
What we learned
This project was a deep dive into the practical applications of large language models. We learned a great deal about prompt engineering, especially how to craft prompts that produce structured JSON output reliably. We also gained significant hands-on experience with the Gemini API, from simple text generation to the more advanced Live API for real-time audio. Architecturally, we learned how to build a powerful, feature-rich application purely on the frontend, which simplifies deployment and reduces infrastructure overhead.
What's next for AI Career Copilot
The future for AI Career Copilot is exciting. We envision several key enhancements: Resume Storage & Versioning: Allow users to create accounts to save their resume and track different versions tailored for various job applications.
Proactive Job Matching: Instead of just analyzing one job, the user could set career goals, and the copilot would proactively scan job boards and alert them to highly-matched opportunities. Deeper Interview Analysis: Incorporate video analysis to provide feedback not just on what a user says, but also on their tone, pacing, and non-verbal cues.
Network Builder: Analyze a user's target industry and suggest professionals to connect with on platforms like LinkedIn, even drafting personalized connection requests.
Log in or sign up for Devpost to join the conversation.