FluentPM — AI PM Interview Coach

Train your voice. Own the room.

An open-source AI interview coach for Product Managers. Practice spoken PM interviews, stakeholder debates, and conversation roleplay with real-time STAR scoring, spaced-repetition vocabulary, and voice I/O — all in the browser.

If FluentPM helps your interview prep, please ⭐ star this repo — it helps others discover it.

---

Demo

📸 Screenshot / GIF coming soon. Run npm run dev to see it live in under 3 minutes.

---

What is FluentPM?

FluentPM is an AI-powered PM interview coach that runs entirely in your browser. It uses DeepSeek (via OpenRouter) for real-time STAR framework scoring, the Web Speech API for voice input and output, and Firebase for session persistence — no backend server required.

Built for Product Managers who need to practice spoken communication under pressure: FAANG interviews, senior PM roles, or everyday stakeholder conversations. Especially useful for non-native English speakers who want structured, AI-coached repetition.

The problem: Generic speaking apps don't simulate PM-specific dynamics — stakeholder pushback, behavioral interview probing, or domain vocabulary like metrics, prioritisation trade-offs, and ownership language.

The solution: FluentPM puts you in realistic PM scenarios with AI opponents that score your STAR-format answers, probe your reasoning, track vocabulary growth, and adapt to your weak points over time.

---

Features

Practice Modes

• Arena — Real-time debate against an AI opponent. Get scored on clarity, structure, and ownership language across multiple turns.
• Interview Simulation — Full PM interview session with an AI interviewer. STAR framework scoring, follow-up probing, and detailed debrief.
• Lightning Round — Rapid-fire Q&A to practice spaced-repetition expressions under time pressure.
• Pushback Drill — Handle tough stakeholder objections with coached responses.
• Quick Drill — 30-second focused answer challenges for busy schedules.
• Podcast Sim — Upload a real conversation transcript, pick your role, tag phrases to practice, and perform your side of the conversation with AI scoring.

Learning System

• Lexicon — Personal expression library with automatic AI enrichment (definition, usage examples, pronunciation tips) and spaced-repetition scheduling.
• Story Bank — Store and manage your STAR stories for behavioral interview prep.
• Custom Questions — Import your own question bank for personalized practice.

Analytics & Gamification

• Stats Dashboard — Session history, difficulty breakdown, scoring trends, and debrief cards.
• Progress Tracker — XP, rank progression, streaks, and achievement badges.
• League — Weekly rankings to add a competitive edge to daily practice.

---

Tech Stack

Category	Technology
Frontend	React 19, Vite 8
Language	JavaScript (ES Modules)
Styling	Inline CSS-in-JS
Database	Firebase Firestore
Auth	Firebase Auth (Google Sign-In)
AI / LLM	OpenRouter API (DeepSeek Chat)
Speech Input	Web Speech API (SpeechRecognition)
Speech Output	Web Speech API (SpeechSynthesis)
Hosting	Firebase Hosting

---

Architecture

User (Google Auth)
       │
       ▼
Firebase Auth ◄──────────────────► Firestore
       │                           (sessions, lexicon,
       ▼                            progress, stories)
  App.jsx (React SPA)
       │
       ├─ Speaking input ──► Web Speech API (STT)
       │
       ├─ AI responses ────► OpenRouter (DeepSeek)
       │                     ↳ System prompts with STAR scoring,
       │                       role isolation, feedback blocks
       │
       ├─ TTS playback ────► Web Speech API (SpeechSynthesis)
       │                     ↳ 12 character voice profiles
       │
       └─ Session save ────► Firestore addDoc

flowchart TD
    A[User speaks] --> B[Web Speech API\nSpeechRecognition]
    B --> C[Transcript text]
    C --> D[OpenRouter API\nDeepSeek Chat]
    D --> E[AI response + score]
    E --> F[Web Speech API\nSpeechSynthesis]
    E --> G[Firestore\nSession saved]
    F --> H[User hears AI]
    G --> I[Stats & Progress\nupdated]

Loading

---

Getting Started

Prerequisites

• Node.js 20+
• A Firebase project with Firestore and Google Authentication enabled
• An OpenRouter API key

1. Clone & Install

git clone https://github.com/ashishworkacc/fluentpm.git
cd fluentpm
npm install

2. Configure Environment

cp .env.example .env

Open .env and fill in your Firebase config values (from Firebase Console → Project Settings → Your apps) and your OpenRouter API key.

3. Start the Dev Server

npm run dev

Open http://localhost:5173 in your browser. Sign in with a Google account to start practicing.

4. Build for Production

npm run build

Deploy the dist/ folder to Firebase Hosting or any static host.

---

Roadmap

• Voice coaching dashboard — WPM, filler word frequency, and pacing trends across sessions
• Podcast Sim history — Compare past podcast sessions and track phrase mastery over time
• Multi-language support — Practice modes adapted for non-English PM interview markets
• AI readiness score — Weekly assessment that surfaces your weakest question types and suggests a daily drill plan

---

Contributing

Contributions are welcome. Please read CONTRIBUTING.md for setup instructions, coding conventions, and the PR process.

---

Legal & Ethics

This project is licensed under the GNU Affero General Public License v3.0 (AGPL-3.0). If you run a modified version as a network service, you must make your source code available to users of that service.

This tool is intended for personal, non-commercial use — language learning, interview preparation, and self-improvement. Please read TERMS_OF_USE.md for the full ethical use guidelines, including attribution requirements and restrictions on commercial redistribution.

For security vulnerabilities, see SECURITY.md.

---

Acknowledgments

• DeepSeek / OpenRouter — LLM backbone
• Firebase — Auth and real-time database
• Web Speech API — Browser-native speech recognition and synthesis
• Vite & React — Build tooling and UI framework