Platform Engineering and DevEx for Your On-Prem LLM
Talkat Platform Engineering Day 2025 - Europe · on Tuesday, 2025-04-01· at 15:20· Level 1 | Hall Entrance N10 | Room H·
summary
Claims a reproducible on‑prem GenAI IDP built on Kubernetes using Helix and an AI spec CRD to version models, knowledge and integrations; it runs open‑source models (Llama 3.3 style), uses GPUs via a GPUuler control plane, and wires RAG pipelines with Haystack and PGVector. Demo shows a working exchange‑rates API integration and a vision RAG that correctly reads a table value (0.35) from a PDF image. Cites urgency (98% of CEOs) and expected productivity gains (20–66%) and argues self‑hosting reduces energy, cost and data‑sovereignty risk.
AI-written summary · about
chapters (12)
- 0:00Self-hosted LLMs needed
- 4:16Introduction
- 6:05Platform engineering role
- 8:35Developer needs with LLMs
- 9:57Developer platform for GenAI
- 10:18Platform stack
- 13:52Testing GenAI apps
- 14:12Exchange rates demo
- 16:18Vision RAG demo
- 18:46GenAI benefits
- 20:54Energy efficiency
- 23:00Summary
related talks
29mBuilding the Bridge Between DevOps and MLOps in a GenAI world
40mEnd-to-end AI Marketplace with Execution Engine on Kubernetes
24mScalable GenAI in Production
49mScaling Data in a Sovereign AI Platform
50mMLOps for a hybrid AI platform compliant with European law
35mBuilding Enterprise Solutions in the AI-Native Era
27mGen-AI at Scale: Simplifying Orchestration of Healthcare Applications Across Multi-Cluster Environme
34mFast-Track Your AI-ML Model Integration