Software 3.0 and the AI Engineering Landscape
summary
Argues that foundation models and hosted model APIs shift ML from bespoke training to product-first workflows so companies build MVPs, collect task data, then specialize models, and that AI+code beats AI+language for building reliable systems. Predicts roughly two orders of magnitude more AI engineers than ML engineers. Lists concrete tooling: Foundation model labs (OpenAI Anthropic StabilityAI), LangChain and LlamaIndex for RAG, vector databases and emerging ops and memory agent stacks, notes JS/TypeScript UI tooling (Vercel, Langflow, Retool) but finds Python dominating core model work.
AI-written summary · about
chapters (17)
- 0:00Introduction and speaker background
- 1:53State of AI engineering overview
- 3:38Foundation models and API shift
- 5:22Defining the AI engineer role
- 7:11GPU arms race and AI growth
- 8:14Foundation model capabilities
- 9:18AI engineering product mindset
- 10:00JavaScript and AI engineering
- 11:05Software 3.0 and AI evolution
- 12:49Code as the key modality
- 14:13LLM application architectures
- 15:18LLM operating systems concept
- 17:04AI stack simplification
- 18:49AI engineering tool ecosystem
- 19:52LLM OS components
- 21:57Domain-specific AI applications
- 23:01Where to start in AI engineering
related talks
28mSoftware 3.0 and The Emerging AI Developer Landscape
19mDeveloper Centric AI
14mWhat Singapore JS devs should know about AI
39mAI and the Democratization of Software Development
26mAl and Software: The Good, The Bad, The Scary
39mForward Focus: Perspectives on AI, Hype, and Security
47mAI and the Future of Software Development
14mFull-Stack AI Engineering from AI-Native Coding to Knowledge, Models, Infrastructure & OSS