Interview Copilot

A real-time desktop-like web app that helps you during technical interviews by:

• Streaming a live (or demo) transcript
• Detecting questions automatically
• Retrieving grounded context from your resume and job description (RAG)
• Generating smart answers in three formats: 15 sec, STAR, Bullets

Built with Go backend + plain HTML/CSS/JS frontend + WebSocket streaming.

---

Quick Start

# 1. Clone / enter the project
cd interview-copilot

# 2. Install Go dependencies
cd backend
go mod tidy

# 3. Run the server (serves frontend automatically)
go run .

# 4. Open in browser
open http://localhost:8080

Requires: Go 1.21+. No CGO required (uses modernc.org/sqlite, pure Go).

---

Usage

Demo Mode (no uploads needed)

1. Open http://localhost:8080
2. Click ▶ Start Demo
3. Watch the transcript fill in, questions get detected, and answers appear in real time.

With Your Own Documents

1. Click 📎 (top right) to open the Knowledge Pack modal
2. Upload your Resume (.txt or .md)
3. Upload the Job Description (.txt or .md)
4. Click ⚡ Index Knowledge Pack
5. Start the demo — answers will now be grounded by your actual documents.

PDF parsing is not implemented in this MVP. Convert to .txt or .md first.

---

Keyboard Shortcuts

Shortcut	Action
⌘1 / Ctrl+1	Switch to 15-second answer
⌘2 / Ctrl+2	Switch to STAR format
⌘3 / Ctrl+3	Switch to Bullets format
⌘K / Ctrl+K	Open command palette
⌘P / Ctrl+P	Pause / resume demo
⌘M / Ctrl+M	Add metrics to answer
⌘F / Ctrl+F	Focus question text box
Esc	Close modal / palette

---

Architecture

interview-copilot/
├── backend/
│   ├── main.go      # HTTP server, REST endpoints, static file serving
│   ├── ws.go        # WebSocket hub + client, command handler
│   ├── db.go        # SQLite init, CRUD for knowledge_chunks
│   ├── schema.go    # Chunk / Snippet types
│   ├── rag.go       # Text chunking, TF keyword scoring, retrieval
│   ├── answer.go    # Mock LLM: generates 15s / STAR / Bullets answers
│   ├── demo.go      # Scripted demo transcript + question/answer pipeline
│   └── go.mod
└── frontend/
    ├── index.html   # 3-column layout, modals, topbar, command bar
    ├── styles.css   # Linear/Notion-like light UI
    └── app.js       # WS client, event handlers, keyboard shortcuts

---

WebSocket Event Schema

All events are JSON with shape { "type": string, "payload": object }.

Event	Payload
session_status	{ live: bool, message: string }
transcript_partial	{ speaker, text, ts }
transcript_final	{ speaker, text, ts }
question_detected	{ question, confidence, tags[], ts }
rag_snippets	{ snippets: [{ source, title, text, score }] }
answer_draft	{ mode: "15s"|"STAR"|"bullets", text }

Client → Server commands

{ "type": "command", "payload": { "name": "shorten" } }
{ "type": "command", "payload": { "name": "add_metrics" } }
{ "type": "command", "payload": { "name": "make_technical" } }
{ "type": "command", "payload": { "name": "clarify" } }
{ "type": "command", "payload": { "name": "switch_mode", "payload": { "mode": "STAR", "question": "..." } } }

---

REST Endpoints

Method	Path	Description
POST	/api/upload?type=resume|jd	Upload file (multipart)
POST	/api/index	Chunk and index uploaded files
GET	/api/status	Indexing status + file list
POST	/api/demo/start	Start demo streaming
POST	/api/demo/stop	Stop demo streaming

---

Replacing the Mock LLM with OpenAI

In backend/answer.go, replace GenerateAnswer() with:

func GenerateAnswer(question string, snippets []Snippet, mode string) string {
    client := openai.NewClient(os.Getenv("OPENAI_API_KEY"))
    // Build prompt from question + snippets + mode instructions
    // Call client.CreateChatCompletion(...)
    // Return response text
}

The rest of the architecture (WebSocket streaming, RAG, UI) stays identical.

---

Sample Knowledge Pack Files

resume.txt

Jane Smith — Senior Software Engineer

Skills: Go, Python, Kubernetes, AWS, PostgreSQL, Redis

Experience:
TechCorp (2021–present): Led backend platform team, built real-time data pipeline handling 1M events/day, reduced deployment time 10x.
StartupXYZ (2019–2021): Full-stack engineer, shipped 3 product features, improved test coverage from 40% to 85%.

Education: BS Computer Science, State University, 2019

jd.txt

Senior Software Engineer — Platform Team

We are looking for a senior engineer to join our platform team.

Requirements:
- 5+ years of software engineering experience
- Strong background in distributed systems
- Experience with Go or Python
- Cloud infrastructure experience (AWS/GCP preferred)
- Excellent communication and collaboration skills

Nice to have:
- Kubernetes experience
- Experience with real-time data systems
- Mentorship experience