Lightning Talk: Harnessing Generative AI Without Making a Faux Pas
summary
Demonstrates a locally hosted FauxPilot setup plus an NVIDIA inference server and FasterTransformer backend to generate complex Kubernetes YAML blueprints (Kanister workflows) from prompts while keeping code local, with the system fine tuned from 11 originals to ~2,000 augmented samples and deployed as a cloud native, serverless app on AWS EC2 using model parallelism for scale; it distinguishes backup workflows for different databases and produces working YAMLs without sending proprietary repositories to external Copilot services.
AI-written summary · about
related talks
06m⚡ Lightning Talk: Supercharge Agentic AI Apps: A DevEx-Driven Approach To Cloud Native Scaffolding
09mLightning Talk: Sailing Kubernetes Operation with AI Power
36mUnleashing Generative AI: Improving Developer Productivity in SRE
09mLightning Talk: Lessons from Putting AI in Front of a Platform: Taming the Non-Dete...
12mCodeless Generative AI Pipelines: Prompt Automation
51mPlatform Engineering in the age of Generative AI
06mLightning Talk: AI-Powered Kubernetes Diagnostics With K8sGPT
48mMastering the AI Shift: Adaptive Strategies for Software Evolution