Podcast
Production AI Engineering: Data Pipelines, Prompt Optimization and Caching
Open original DataTalks.Club episode
Production AI Engineering: Data Pipelines, Prompt Optimization and Caching
Original Episode
Use these links for the canonical episode and media sources.
- Open the original DataTalks.Club podcast page
- Watch on YouTube
- Listen on Spotify
- Listen on Apple Podcasts
Episode Overview
How do you move AI projects from proof-of-concept to reliable production systems while keeping prompts, pipelines, and response times under control? In this episode Bartosz Mikulski, an AI and data engineer who specializes in productionizing AI, breaks down the engineering work required to make models dependable beyond demos. Bartosz explains how to design robust data pipelines, apply prompt optimization practices, and introduce caching strategies that reduce load and improve responsiveness. He also covers.
People
Use these links to connect the episode to guest notes.
Chapter Summary
Use these checkpoints to decide whether to open the source transcript.
- 0:00 - Episode Opening & Guest Overview (Data Intensive AI)
- 2:02 - Book Contribution Clarified & Testing Focus
- 4:00 - Career Path: Java → Data Engineering → AI Engineering
- 6:04 - Publishing Routine: Blogging Frequency & Content Practice
- 9:05 - Data Trust: Why Testing Prevents “This Number Doesn’‘t Look Correct”
- 11:47 - Test Strategy for Data Pipelines: Snapshot & Integration Testing
- 13:14 - Testing Tools: Great Expectations, Soda, SQL Tests vs Spark Tests
- 17:10 - Technology Choice: When to Use Apache Spark
- 18:38 - Data Engineering’s Role in AI: Preprocessing & Fine-Tuning Data
- 21:46 - Invisible AI Use Cases: Augmented Generation & Review Analysis
- 25:13 - Prompt Engineering Basics: In-Context Learning & Examples
- 28:16 - Prompt Evaluation: Formatting, Examples, and Cost Tradeoffs
- 30:00 - Prompt Compression: Token Optimization Techniques
- 31:45 - Prompt Caching & Model Efficiency (attention caching, Claude)
- 33:42 - Open-Source Models & Tools Experience (DeepSeek, Perplexity)
- 35:54 - AI for Lead Scoring: LinkedIn Automation & Qualification
- 41:04 - Chrome Extension Architecture: Backend AI Integration Pattern
- 42:05 - Coding Assistants: Cursor Workflow & Productivity Boosts
- 44:38 - Code AI Comparison: Cursor vs GitHub Copilot & Alternatives
- 47:19 - Search-Focused Assistants: Using Perplexity & Tool Selection
- 52:09 - Website Hosting: Static Site Generators & GitHub Pages
- 53:10 - Blogging as Business: Attracting Clients & Teaching Workshops
- 56:17 - AI-Assisted Writing: Drafting, Rewriting, and Maintaining Voice
- 1:00:21 - Episode Wrap-Up & Guest Resources (blog link invitation)