What is mattwood.fyi?

A live, structured feed of riffs and links from Matt Wood, Chief AI & Technology Officer at AWS — readable by people and queryable by agents as a typed knowledge graph.

Is there an API and is it free?

Yes. A public, read-only query API is available with no authentication and no API keys. See https://mattwood.fyi/openapi.json and https://mattwood.fyi/docs.

How should an AI agent use this site?

Fetch https://mattwood.fyi/feed.json for the full structured feed, or query the API for keyword/semantic search and the connection graph. See https://mattwood.fyi/agents/ for when-to-use guidance.

Real-Time Video Generation on Trainium

Connections

Supports

Inside the Inference Hardware Revolution Of 2026 Both examine specialized hardware acceleration for AI inference; the Trainium kernel optimization work is a concrete case study within the broader inference hardware revolution documented in 2026
Mini-AGI: Continual Learning on 8GB VRAM Both address making computationally intensive AI workloads feasible on constrained or specific hardware targets through optimization, suggesting a broader trend of democratizing high-performance AI inference

Related to

Cognition and AWS Strategic Partnership for Autonomous Engineers AWS Trainium is an Amazon product; the Cognition-AWS partnership for autonomous engineers reflects Amazon's broader push into AI infrastructure, of which Trainium hardware optimization is a key technical component
Training a 4B Model for Faster SQL Query Plans Both focus on kernel-level and hardware-specific optimizations to accelerate AI model execution — one for video generation on Trainium, the other for SQL query planning on a 4B model
ARGODRIVE Deltafin: Streaming Large MoE Models on Apple Silicon Both tackle the challenge of running demanding generative AI workloads on specific hardware accelerators (Trainium vs Apple Silicon) through low-level optimization techniques like kernel tuning and model streaming

Challenges

The Plunging Price of Thought Kernel optimization for real-time video generation on specialized hardware demonstrates that inference costs require significant engineering effort to reduce, nuancing claims that AI thought is simply becoming cheap

Type:	link
First published:	Saturday, 26 September 2026 at 10:35 UTC
Location:	https://www.amazon.science/blog/a-kernel-centric-path-to-real-time-video-generation-on-trainium?utm_campaign=a-kernel-centric-path-to-real-time-video-gener&utm_medium=organic-asw&utm_source=linkedin&utm_content=2026-09-25-a-kernel-centric-path-to-real-time-video-gener&utm_term=2026-september
URL:	https://mattwood.fyi/i/5838311e

I'm Matt Wood, and this is For Your Information. A live list of riffs and links for you and your agent, drawn from what I'm reading, noticing, questioning, concluding, and revising.

Connections