Tech Deep Dive - Daily Podcast Agent

A self-running agent that generates a daily 1-hour podcast episode, each covering one open-source technology in depth. It discovers the best YouTube conference talks and tutorials, transcribes and summarizes them with AI, and produces a professional audio episode you can listen to on your phone.

How It Works

1. Picks today's topic from a curated curriculum (Hadoop, Spark, Kafka, PostgreSQL, ClickHouse, etc.)
2. Searches YouTube for the best videos on that technology (conference talks, tutorials, deep dives - any date)
3. Extracts transcripts from the top videos
4. Summarizes with AI (Google Gemini) - extracting architecture details, use cases, and key insights
5. Writes a podcast script with segments: Intro → What & Why → Architecture → Use Cases → Comparisons → Outro
6. Generates audio using edge-tts (Microsoft Neural voices)
7. Serves a web app with a mobile-friendly podcast player (PWA - installable on your phone)

Deploy Free (No Install on Your Computer)

You need two free API keys first (one-time, takes 2 minutes):

1. YouTube Data API key: Google Cloud Console → Create project → Enable "YouTube Data API v3" → Credentials → Create API Key
2. Gemini API key: Google AI Studio → Create API Key

Then deploy to the cloud for free:

Deploy to Render.com (Recommended)

1. Push this repo to GitHub (or fork it)
2. Go to render.com → Sign up free (no credit card)
3. Click New → Web Service → Connect your GitHub repo
4. Render auto-detects the Dockerfile. Set these:
   • Name: tech-deep-dive
   • Plan: Free
5. Add Environment Variables:
   • YOUTUBE_API_KEY = your key
   • GEMINI_API_KEY = your key
6. Click Deploy

You'll get a permanent URL like https://tech-deep-dive.onrender.com. Open it on your phone, tap Share → "Add to Home Screen" to install it as an app.

Note: Render free tier sleeps after 15 min of inactivity. First visit after sleep takes ~30 seconds to wake up, then it's fast.

---

Alternative: Run Locally

1. Prerequisites

• Python 3.11+
• ffmpeg: brew install ffmpeg

2. Install & Run

cd tech-podcast-agent
python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

cp .env.example .env
# Edit .env and paste your API keys

python run.py

Open the URL shown in the terminal on your phone (same WiFi). Tap Share → "Add to Home Screen" to install as an app.

CLI Usage

# Show upcoming curriculum schedule
python -m src.main --schedule

# Generate episode for a specific technology
python -m src.main --topic "Apache Spark"

# Jump to day N of the curriculum
python -m src.main --day 5

Curriculum

The agent covers 30+ technologies, one per day. Edit curriculum.yaml to customize:

Day	Technology	Category
1	PostgreSQL	Databases
2	Apache Cassandra	Databases
3	MongoDB	Databases
4	Apache Hadoop	Big Data
5	Apache Spark	Big Data
6	Apache Flink	Stream Processing
7	Apache Kafka	Messaging
8	Elasticsearch	Search
9	OpenSearch	Search
10	ClickHouse	Analytics
11	DuckDB	Analytics
...	...	...

After completing all topics, the cycle repeats with fresh content.

Episode Structure

Each ~60-minute episode follows this format:

• Intro (~3 min) - Welcome and topic introduction
• What & Why (~10 min) - Fundamentals, history, problem it solves
• Architecture (~20 min) - Deep dive into internals and design
• Use Cases (~15 min) - Real-world production deployments
• Comparisons (~7 min) - How it stacks up against alternatives
• Outro (~5 min) - Key takeaways and tomorrow's preview

Configuration

Edit config.yaml to customize:

• TTS voice and speed
• Target podcast duration
• Web server port
• YouTube search parameters

Edit curriculum.yaml to:

• Add/remove/reorder technologies
• Customize search queries per technology

Cost

$0/month - all services used are within free tier limits:

• YouTube Data API: 10,000 units/day (uses ~500)
• Gemini 1.5 Flash: 1,500 requests/day (uses ~15)
• edge-tts: unlimited, no API key needed
• Render.com hosting: free tier (no credit card)
• edge-tts: unlimited, no API key needed

Project Structure

tech-podcast-agent/
├── run.py                 # Start the web app
├── config.yaml            # Settings
├── curriculum.yaml        # Technology learning schedule
├── requirements.txt       # Python dependencies
├── src/
│   ├── main.py            # Pipeline orchestrator
│   ├── web.py             # FastAPI web app
│   ├── curriculum.py      # Curriculum & progress tracking
│   ├── discovery.py       # YouTube content discovery
│   ├── transcriber.py     # Transcript extraction
│   ├── summarizer.py      # Gemini AI summarization
│   ├── script_writer.py   # Podcast script generation
│   ├── audio_generator.py # edge-tts audio generation
│   ├── podcast_assembler.py # Final MP3 assembly
│   └── utils.py           # Shared utilities
├── templates/
│   └── index.html         # Web app UI
├── static/                # CSS, JS, icons
├── output/                # Generated podcast MP3s
└── data/                  # Progress tracking