Coding Trainer

An AI-powered coding review trainer that dynamically generates programming challenges using the Anthropic API (Claude). Each session produces fresh challenges across 5 structured learning blocks and 5 programming languages.

Phase 1 — FastAPI + Railway/Render deployment (this repo)
Phase 2 — AWS Lambda + API Gateway + S3 rebuild (coming after Phase 1 is live)

---

What It Does

Select a programming language, pick a learning block, and Claude generates a tailored challenge on demand. Every session is different — same learning objectives, fresh scenario each time — so the app has genuine replay value.

5 Learning Blocks

#	Block	What You Practice
1	Language Fundamentals	Core syntax and idiomatic patterns
2	Code Review	Spot bugs, security flaws, and inefficiencies
3	Algorithm Patterns	Hashmaps, stacks, sliding window, recursion
4	Edge Cases & Error Handling	Defensive coding and robust error handling
5	Full Module Review	Correctness, efficiency, security, readability

5 Languages: Python, JavaScript, C#, Java, Go

---

Architecture (2-Tier + External API)

┌────────────────────────────────────────────────────────────┐
│  Presentation Tier                                         │
│  HTML / JS / CSS — single-page frontend                    │
│  Language selector → block nav → challenge → hint/answer   │
└────────────────────────────┬───────────────────────────────┘
                             │  POST /generate
                             ▼
┌────────────────────────────────────────────────────────────┐
│  Application Tier                                          │
│  FastAPI (Python) — main.py                                │
│  • Validates request (language, block, difficulty)         │
│  • Builds structured prompt for each block type            │
│  • Calls Anthropic API via anthropic Python SDK            │
│  • Returns structured JSON challenge                       │
└────────────────────────────┬───────────────────────────────┘
                             │  External API call
                             ▼
                  ┌──────────────────────┐
                  │  Anthropic API       │
                  │  (external service)  │
                  │  claude-haiku-4-5    │
                  └──────────────────────┘

No data tier — there is no database or file store. The Anthropic API is a third-party external service, not an owned data layer. A true data tier would be added in a future version (e.g., storing user session history).

API endpoint: POST /generate

Request:

{ "language": "python", "block_number": 1 }

Response:

{
  "challenge": "...",
  "context":   "...",
  "hint":      "...",
  "answer":    "...",
  "tags":      ["list-comprehension", "iteration"],
  "difficulty": "beginner"
}

Valid languages: python, javascript, csharp, java, go
Valid block numbers: 1 through 5

---

Local Setup

Prerequisites: Python 3.11+ and an Anthropic API key

# 1. Clone the repo
git clone https://github.com/your-username/coding-trainer.git
cd coding-trainer

# 2. Install uv if you don't have it
curl -LsSf https://astral.sh/uv/install.sh | sh  # macOS/Linux
# or: winget install astral-sh.uv                 # Windows

# 3. Install dependencies and create virtual environment
uv sync

# 4. Set up your API key
copy .env.example .env        # Windows
cp .env.example .env          # macOS / Linux
# Then open .env and replace "your_key_here" with your actual key

# 5. Run the server
uv run uvicorn main:app --reload --port 8000

Open http://localhost:8000 in your browser.

Note: Use uvicorn directly for local development — not the Procfile (which uses $PORT, a Unix env var).

---

Cloud Deployment

Railway

1. Push this repo to GitHub
2. Create a new project at railway.app → Deploy from GitHub repo
3. In the Railway project settings → Variables → add:

   ANTHROPIC_API_KEY = sk-ant-your-key-here
   

4. Railway auto-detects the Procfile and deploys. Your app will be live at a *.up.railway.app URL.

Render

1. Push this repo to GitHub
2. Create a new Web Service at render.com
3. Connect your GitHub repo
4. Settings:
   • Build Command: pip install -r requirements.txt
   • Start Command: uvicorn main:app --host 0.0.0.0 --port $PORT
5. In Environment → add:

   ANTHROPIC_API_KEY = sk-ant-your-key-here
   

6. Deploy. Render provides a *.onrender.com URL.

Cold starts: Free-tier instances on both platforms sleep after inactivity. The first request after idle may take 10–15 seconds — the frontend shows a "Server warming up" message automatically.

---

Fork & Deploy Your Own Instance

This project is designed to be self-hosted. Every fork requires its own Anthropic API key — no shared keys, no shared costs.

1. Fork this repo on GitHub
2. Deploy to Railway or Render (steps above)
3. Add your own ANTHROPIC_API_KEY in the platform's environment settings
4. Done — your own private instance, your own costs

---

Extending the App

Add a new language:

1. Add it to LANGUAGE_DISPLAY in main.py
2. Update the validate_language allowed set
3. Add a <button> to the language selector in static/index.html

Add a new block:

1. Add a prompt to the prompts dict in build_prompt() in main.py
2. Update the block selector in static/index.html
3. Update BLOCK_NAMES in the JavaScript

Switch to a more capable model:
In main.py, change model="claude-haiku-4-5-20251001" to "claude-sonnet-4-6" for higher quality challenges at higher cost.

---

Screenshots

Coming soon — add screenshots of the challenge card, hint reveal, and answer reveal here.

---

Phase 2 — AWS Rebuild (Planned)

After Phase 1 is live, this project will be rebuilt on AWS infrastructure to demonstrate cloud architecture skills:

• AWS Lambda — Python function handling challenge generation
• API Gateway — HTTP endpoint in front of Lambda
• S3 — Static frontend hosting
• Secrets Manager / SSM — API key storage (no .env files)

Both versions will remain available as separate branches/repos. The two-phase story (fast deployment → cloud-native rebuild) demonstrates both product delivery and AWS cert knowledge.

---

License

MIT — see LICENSE