Gaming Roundtable
- Matt: Vibe coding.
- Brent: Vibe coding, WoW Forever, and Ocarina of Time.
- Brandon: Path of Exile 2 and WoW Forever.
Meet Dominic, an AI Engineer: The Job and How to Get In
We're joined by AI engineer Dominic for a full-episode interview. He walks us through what an AI engineer does day to day, how he got there, and how the role differs from a developer or a data scientist. Then we get practical: will AI replace programmers or make them far more productive, are entry-level tech jobs disappearing, and where should someone in an unrelated career start? We cover whether you need a degree, how much math it really takes, and what belongs in a beginner's portfolio.
Will AI Replace Software Engineers?
We dig into the "SaaSpocalypse": is AI-built software a real threat to software vendors, or an overreaction? Could a healthcare organization replace its EMR, or a midsize business its accounting system, with software built with AI, even with no developers on staff? We talk through why writing the code is the easy part compared to maintaining, securing, and supporting it, and why something you can build in a weekend still isn't ready for 10,000 employees.
Then the people question: how worried should ordinary people be, what does "losing control of AI" actually mean in engineering terms, and how do today's real harms (fraud, cyberattacks, misinformation, and jobs) compare with the speculative ones?
Agents and Architecture
Why the setup around an AI agent matters more than the model itself. We cover how an agent differs from chatting with an LLM, what files like AGENTS.md and CLAUDE.md are for (they set conventions; they aren't security boundaries), one powerful agent versus several specialized ones, keeping token costs down, and the most common mistakes people make.
Securing AI Before It Breaks Something
The risks of running a coding agent on your own computer, from stolen tokens and hijacked sessions to the "oops, I deleted your database" moment. We look at how prompt injection works and why you can't just tell the AI to ignore it, the minimum safe setup for hobbyists (containers or VMs, least privilege, and careful API key handling), and how to give an agent autonomy while keeping a human in the loop for high-risk actions. Our case study is the 2025 Replit incident, in which an AI agent deleted a production database, and the guardrails that would have limited the damage.
Five-Year Predictions
We close with Dominic's boldest calls: what AI will be doing in five years that people won't believe today, what excites and worries him, the biggest misconception about AI, and what an ordinary person should do right now.
Watch & Listen
VOD: https://youtu.be/g24Kiojs518
Twitch: https://twitch.tv/TheDigitalAscent
YouTube: https://youtube.com/@TheDigitalAscent
Spotify: https://creators.spotify.com/pod/profile/digitalascent/
Hosts & Credits
- Hosts: Brent, Matt, Brandon
- Guest: Dominic (AI engineer)
- Technical Design / Production: Brent