Developer Centric AI
summary
Argues that developer-centric AI now means building on foundation models with practical tooling: stable APIs compatible with OpenAI, PyTorch at scale for deployment, hosted families like Falcon and Llama2, and easy fine-tune recipes to adapt models to private data. Claims reduced data needs via fine-tuning, offers performance wins — 10× latency reduction and 11× throughput from model-system co-design — and pricing advantages (example: 5× cheaper base fine-tunes, 30× on tuned models). Integrates with LangChain and major clouds for fast product experiments to production.
AI-written summary · about
chapters (10)
- 0:00Developer-centric AI intro
- 1:10Research to generative AI
- 3:20Research to production challenges
- 6:35Navigating AI model landscape
- 9:08Choosing model and approach
- 11:58Opportunities in AI ecosystem
- 13:46Simplifying AI with Fireworks
- 14:08Optimizing AI models
- 15:57Building Gen AI products
- 17:22Future of developer-centric AI
related talks
24mSoftware 3.0 and the AI Engineering Landscape
46mThe AI-Powered Developer
47mEmbedding AI: Don’t Burn Your Fingers in the Heat of the AI revolution
28mTransforming Internal Developer Platforms with AI: Accelerating Innovation
18mLeveraging AI to Improve the Developer Experience
30mDeveloper Productivity for GenAI Apps
05mOpening Remarks
06mThe Building Blocks of Agentic Al