Skip to content

RaghavSethi006/CareerCo-Pilot

BranchesTags
 
 

Repository files navigation

Hacked 2026 Team Career Savers Status

🚀 Career Co-Pilot

An AI-Powered Career Intelligence & Application Optimization System
Maximize interview conversions — not application volume.

📖 Table of Contents

🎯 The Problem

Job hunting is broken. Students and early-career professionals face:

Pain Point Impact
🔄 Spray-and-pray applications 300+ applications with < 5% response rates
📄 Generic resumes One-size-fits-all resumes that ATS systems filter out
🕵️ Scattered job sources Hours wasted manually searching across dozens of platforms
🤖 Spam auto-apply bots Mass applications that get accounts flagged and banned
📊 Zero feedback loops No way to learn what's working and what isn't

The goal is not to apply to 300 jobs. The goal is to maximize interview callbacks per 10 applications.

💡 Our Solution

Career Co-Pilot is a local-first, AI-powered career intelligence system that transforms how you search, apply, and prepare for jobs. Instead of automating volume, we optimize for quality.

  Prepare  →  Review  →  Approve  →  Submit
     ↑                                  │
     └──── Feedback Learning Loop ──────┘

What makes us different?

Feature Spam Bots Career Co-Pilot
Application control Mass unattended User approves each one
Resume generation Generic template AI-tailored per job description
Submission Invisible, headless Visible browser — you see every click
Data privacy Cloud-scraped 100% local-first
Intelligence None Learns from your outcomes over time
Interview prep ❌ Auto-generated kits per role
Compliance ⚠️ Risky Ethical, rate-limited, user-controlled

✨ Key Features

📋 Career Profile Engine

Build a structured career knowledge graph from your skills, projects, experience, and certifications. The system infers skill relationships (e.g., REST APIs → HTTP knowledge), calculates confidence scores, and dynamically clusters your abilities.

🔍 Hybrid Job Discovery Engine

Intelligently discovers jobs from safe, permissible sources using recursive intent expansion — not blind crawling. Generates smart search queries from your profile, deduplicates results with hash + semantic similarity matching, and ranks every opportunity.

📊 Intelligent Job Ranking

Every discovered job receives a composite Interview Probability Score based on:

  • Skill match score
  • Experience alignment
  • Role depth analysis
  • Skill gap penalty
  • Seniority compatibility

📝 AI Resume Generation (Gemini-Powered)

A full pipeline that parses job descriptions, selects the most relevant fragments from your profile, rewrites bullets contextually with strict anti-hallucination guardrails, optimizes for ATS keyword density, and produces a tailored resume — with explainability for every included bullet.

📨 Application Draft Engine

Prepares complete application packages — tailored resume, cover letter, and pre-filled screening question answers — without ever submitting. Everything stays in draft until you explicitly approve.

🖥️ Controlled Submission Engine

After your approval, a visible browser session opens and fills out the application form in real-time. You watch every field get filled. A double-approval flow means you see the completed form before the final submit click. Safety features include daily submission caps, human-like timing delays, and zero unattended submissions.

🎯 Interview Preparation Engine

Automatically generates targeted interview kits when an application moves to the interview stage:

  • Technical questions calibrated to the job description
  • Behavioral prompts with STAR-format answer drafts
  • Company-specific deep-dive questions
  • Weakness-targeting questions based on your skill gaps
  • Difficulty scaling across Easy → Medium → Hard

📈 Feedback Learning Loop

The system learns from your outcomes — tracking approved/rejected applications, interview callbacks, and offer rates — to continuously adapt ranking weights, resume tone, role priorities, and skill importance. Cold-start guards ensure no unreliable patterns surface before sufficient data.

📊 Insights & Analytics Dashboard

Real-time metrics with interactive Recharts visualizations:

  • Application volume over time
  • Response rate & interview rate trends
  • Match distribution breakdowns
  • Best-performing resume versions
  • Top missing skills across rejections

🏗️ System Architecture

┌─────────────────────────────────────────────────────────────┐
│                     Desktop UI (React + Vite)               │
│  ┌─────────┐ ┌────────────┐ ┌──────────┐ ┌──────────────┐  │
│  │  Jobs    │ │Applications│ │  Resume  │ │  Interviews  │  │
│  │ Library  │ │  Kanban    │ │  Studio  │ │  Prep Kit    │  │
│  └─────────┘ └────────────┘ └──────────┘ └──────────────┘  │
│  ┌─────────┐ ┌────────────┐ ┌──────────┐                   │
│  │ Profile  │ │  Insights  │ │ Settings │                   │
│  │ Editor   │ │ Dashboard  │ │   Page   │                   │
│  └─────────┘ └────────────┘ └──────────┘                   │
└──────────────────────┬──────────────────────────────────────┘
                       │  REST API
┌──────────────────────▼──────────────────────────────────────┐
│               Local Backend (FastAPI + Python)              │
│  ┌──────────────────────────────────────────────────────┐   │
│  │                  AI Engine Layer                      │   │
│  │  ┌─────────┐ ┌────────┐ ┌──────────┐ ┌───────────┐  │   │
│  │  │Discovery│ │ Resume │ │ Draft &  │ │ Interview │  │   │
│  │  │ Engine  │ │ Engine │ │Submission│ │  Engine   │  │   │
│  │  └─────────┘ └────────┘ └──────────┘ └───────────┘  │   │
│  │  ┌─────────┐ ┌────────┐                              │   │
│  │  │Feedback │ │Profile │                              │   │
│  │  │ Loop    │ │ Engine │                              │   │
│  │  └─────────┘ └────────┘                              │   │
│  └──────────────────────────────────────────────────────┘   │
│  ┌────────────┐  ┌─────────────┐  ┌────────────────────┐   │
│  │ Gemini AI  │  │  Playwright │  │   APScheduler /    │   │
│  │  Client    │  │  Browser CDP│  │   Async Scheduler  │   │
│  └────────────┘  └─────────────┘  └────────────────────┘   │
└──────────────────────┬──────────────────────────────────────┘
                       │
┌──────────────────────▼──────────────────────────────────────┐
│                Local Storage Layer                          │
│  ┌────────────────────────────────────────────────────────┐ │
│  │                    SQLite Database                     │ │
│  │  user_profile │ jobs │ resume_versions │ app_drafts   │ │
│  │  interview_kits │ insights_cache │ discovery_runs     │ │
│  └────────────────────────────────────────────────────────┘ │
└─────────────────────────────────────────────────────────────┘

🧠 AI & Intelligence Pipeline

graph LR
    A[User Profile] --> B[Role Inference Engine]
    B --> C[Query Generator]
    C --> D[Job Discovery<br/>Remotive API / RSS]
    D --> E[Normalizer + Dedup]
    E --> F[Ranking Engine]
    F --> G[Job Library UI]

    G --> H[Resume Engine]
    H --> H1[JD Analyzer]
    H1 --> H2[Fragment Selector]
    H2 --> H3[Bullet Rewriter<br/>Gemini AI]
    H3 --> H4[Resume Compiler]

    H4 --> I[Draft Engine]
    I --> I1[Form Analyzer<br/>Playwright]
    I1 --> I2[Draft Generator]
    I2 --> J[Review Dashboard]

    J -->|Approve| K[Submission Engine<br/>Visible Browser]
    K --> L[Outcome Tracking]
    L --> M[Feedback Loop]
    M --> N[Weight Adapter]
    N -->|Improved Rankings| F

    J -->|Interview| O[Interview Kit<br/>Generator]
    O --> P[Mock Simulator]
Loading

Anti-Hallucination Guardrails

Our AI resume generation follows strict rules to ensure zero fabricated claims:

  • ✅ Preserve ALL factual claims exactly as written
  • ✅ Never add metrics, numbers, or technologies not in the original
  • ✅ Graceful fallback to rule-based output when Gemini is unavailable
  • ✅ Bullet length checks (revert to original if rewrite is > 50% longer)
  • ✅ Every included fragment is linked to a specific job requirement (explainability)

🖥️ Tech Stack

Frontend

Technology Purpose
React 18 Component-based UI framework
Vite 5 Lightning-fast build tool & dev server
TypeScript 5.4 Type-safe development across all packages
TailwindCSS 3 Utility-first styling with custom design tokens
Zustand Lightweight state management with localStorage persistence
React Router v6 Client-side routing for all 7 feature pages
@dnd-kit Accessible drag-and-drop for the Kanban board
Recharts Composable charting library for insights dashboard

Backend

Technology Purpose
Python 3.11+ Core backend language
FastAPI 0.110 High-performance async API framework with auto-docs
Google Gemini 1.5 Flash AI model for resume tailoring & interview generation
Playwright Browser automation for form analysis & controlled submission
SQLite Zero-config local-first database
pypdf + python-docx Document parsing for resume upload
Uvicorn ASGI server with hot-reload for development

Infrastructure

Technology Purpose
Docker Compose One-command deployment of frontend + backend
Nginx Reverse proxy for production frontend
pnpm Workspaces Monorepo package management
Chrome DevTools Protocol CDP-based browser automation for visible submissions

📦 Monorepo Structure

Career-Co-Pilot/
├── 📁 packages/                    # Frontend monorepo (pnpm workspaces)
│   ├── 📦 core/                    # Shared TypeScript interfaces & domain types
│   ├── 📦 api/                     # Mock data layer & API client functions
│   ├── 📦 ui/                      # Shared design system (Sidebar, MatchBadge, StatusPill, SplitPane)
│   ├── 📦 app/                     # Vite entry point, router, AppShell, feature registry
│   ├── 📦 feature-jobs/            # Job Intelligence Library — browse, search, filter
│   ├── 📦 feature-applications/    # Kanban board — drag-and-drop application tracking
│   ├── 📦 feature-resume-studio/   # Split-pane resume editor — base vs tailored comparison
│   ├── 📦 feature-interviews/      # 4-tab interview prep — questions, STAR drafts, mock sim
│   ├── 📦 feature-insights/        # Analytics dashboard — charts, metrics, trend indicators
│   ├── 📦 feature-profile/         # Career profile viewer — skills, experience, projects
│   └── 📦 feature-settings/        # App settings — LLM config, discovery, backup/restore
│
├── 📁 backend/                     # Python FastAPI backend
│   ├── 📁 app/
│   │   ├── 📁 engines/
│   │   │   ├── 📁 discovery/       # Job discovery — query gen, adapters, dedup, ranking
│   │   │   ├── 📁 resume/          # Resume generation — JD analyzer, fragment selector, compiler
│   │   │   ├── 📁 applications/    # Draft engine & controlled submission
│   │   │   ├── 📁 interviews/      # Interview kit generator
│   │   │   ├── 📁 feedback/        # Metrics aggregator, pattern detector, weight adapter
│   │   │   └── 📁 profile/         # Career profile engine
│   │   ├── 📁 routers/             # 12 FastAPI routers (REST API endpoints)
│   │   ├── 📁 clients/             # External API clients (Gemini)
│   │   ├── 📁 db/                  # SQLite schema & data access layer
│   │   ├── 📁 data/                # Static data files (skill taxonomy, role synonyms)
│   │   ├── main.py                 # FastAPI app entry point
│   │   └── scheduler.py            # Background job discovery scheduler
│   ├── 📁 tests/                   # Backend test suite
│   ├── Dockerfile
│   └── requirements.txt
│
├── 📁 docs/                        # Documentation & team briefs
│   └── 📁 team/                    # Individual engineer assignment docs
│
├── docker-compose.yml              # Full-stack Docker orchestration
├── pnpm-workspace.yaml             # Monorepo workspace config
├── tsconfig.base.json              # Shared TypeScript config
└── package.json                    # Root scripts (dev, build, typecheck)

🐳 Docker Deployment

# From repo root
docker compose up --build
# → frontend: http://localhost:3000
# → backend:  http://localhost:8000
Service URL
Frontend http://localhost:3000
Backend API http://localhost:8000
Swagger Docs http://localhost:8000/docs
Health Check http://localhost:8000/health 
# Tear down
docker compose down

🌐 Visible Browser Setup

For AI Auto Search and Assisted Apply features, Career Co-Pilot connects to a Chrome instance via the Chrome DevTools Protocol. Start Chrome with remote debugging before using browser-assisted features.

macOS 
open -na "Google Chrome" --args --remote-debugging-port=9222 --remote-debugging-address=0.0.0.0 --user-data-dir=/tmp/career-copilot-cdp
Windows (PowerShell or CMD) 
# If Chrome is in PATH:
chrome.exe --remote-debugging-port=9222 --remote-debugging-address=0.0.0.0 --user-data-dir=%TEMP%\career-copilot-cdp

# If Chrome is not in PATH:
"C:\Program Files\Google\Chrome\Application\chrome.exe" --remote-debugging-port=9222 --remote-debugging-address=0.0.0.0 --user-data-dir=%TEMP%\career-copilot-cdp

Verify Chrome debug endpoint:

curl http://localhost:9222/json/version

🔒 The browser session is fully visible — you see every action Career Co-Pilot performs. Nothing happens behind a headless curtain.

🔌 API Reference

Career Co-Pilot exposes 12 RESTful routers through FastAPI with full OpenAPI documentation at /docs.

Endpoint Group Key Routes Description
Health GET /health, GET /ready Service health & readiness checks
Profile GET /profile, PUT /profile Career profile CRUD
Jobs GET /jobs, GET /jobs/:id, DELETE /jobs/:id Job library management
Discovery POST /discovery/run, GET /discovery/status Trigger & monitor job discovery runs
Resumes POST /resumes/generate, GET /resumes, GET /resumes/:id AI resume generation & management
Drafts POST /drafts/prepare, PATCH /drafts/:id, POST /drafts/:id/approve Application draft lifecycle
Applications GET /applications, PATCH /applications/:id/status Application tracking
Interviews POST /interviews/:id/generate, GET /interviews/:id Interview kit generation
Insights GET /insights, POST /insights/refresh, GET /insights/patterns Analytics & feedback patterns
Outcomes PATCH /applications/:id/outcome Track application results
Settings GET /settings, PUT /settings App configuration

🛡️ Security & Compliance

Career Co-Pilot is built with ethical automation at its core:

Principle Implementation
🏠 Local-First All data stays on your machine — no cloud harvesting
✋ User-Controlled Every application requires explicit human approval
🔒 Double-Approval Submission You see the filled form before the final submit click
🚦 Rate-Limited Configurable daily submission cap (default: 10/day)
🐢 Human-Like Timing Random 300–1200ms delays between form field fills
🚫 No Proxy Rotation No CAPTCHA bypassing, no fake accounts
👁️ Visible Browser Non-headless — you watch every automated action
📊 Audit Trail Every discovery run, draft, and submission is logged

Career Co-Pilot is an AI-assisted application preparation tool — not a mass application bot.

🗺️ Roadmap

Phase Features Status
Phase 1 Career Profile Engine, Resume Generator, Role Inference ✅ Complete
Phase 2 Hybrid Job Discovery, Ranking Engine, Skill Gap Analysis ✅ Complete
Phase 3 Application Draft Engine, Review Dashboard, Controlled Submission 🔨 In Progress
Phase 4 Interview Engine, Feedback Learning System 🔨 In Progress
Phase 5 Tauri Desktop App, Local Embedding Store, Advanced Analytics 📋 Planned

👥 Team — Career Savers

We are a team of 5 engineers from Hacked 2026, each owning distinct verticals across the full stack:

Name Role Frontend Ownership Backend Ownership
Sashreek 🔗 Integration Owner core types, api mock data, app wiring & routing SQLite schema, data layer, profile/jobs/applications routers
Ajax 🎮 Interaction Engineer feature-applications Kanban, feature-resume-studio editor Application Draft Engine, Controlled Submission Engine
Haq 📊 Data Visualization feature-interviews prep, feature-insights analytics Resume Generation Engine (Gemini AI), Interview Kit Generator
Divy ⚙️ Infrastructure & Settings feature-settings page Job Discovery Engine, Scheduling, Docker infrastructure
Raghav 🎨 Design System & Profile ui shared components, feature-profile page Feedback Learning Loop, Insights Analytics Engine

🏆 Built for Hacked 2026

Career Co-Pilot — because your next career move deserves strategy, not spam.

Built with ❤️ by Team Career Savers

📄 License

This project is licensed under the Career Co-Pilot Protective Open Source License (CC-POL v1.0). See LICENSE.

About

No description, website, or topics provided.

Resources

Readme

License

View license
Activity

Stars

0 stars

Watchers

0 watching

Forks

0 forks
Report repository

Releases

No releases published

Packages

 
 
 

Contributors

Languages

  • Python 57.6%
  • TypeScript 41.3%
  • TeX 0.7%
  • JavaScript 0.2%
  • CSS 0.1%
  • Dockerfile 0.1%