Writing

Articles published across the web.

11 entries

Building with Jev: use cases and implementation architecturesBlogUsing fast decision models in search, browser agents, model routing, and other applications.

Transformer basics: prefill, decode, and the KV cacheBlogHow a prompt moves through embeddings, attention, prefill, and token-by-token decoding.

Jev and System One modelsBlogShared prefill, decision training, and fast judgments inside software.

Multi-agent systems scale thought in parallelBlogWhen parallel agents save time, and how coordination and duplicated work limit the gains.

Compound Engineering: My Self-Improving Remote Agent SetupBlogA Slack and Hermes setup where conversations, corrections, and completed tasks improve future work.

Durable execution vs atomic executionBlogA blog post comparing durable and atomic execution strategies.

Speculative decodingBlogA knowledge post on speculative decoding, a technique to speed up LLM inference by using a smaller model to generate candidate tokens and a larger model to verify them.

Vision-Action Transformers vs. VLA vs. World-Action ModelsBlogA blog post comparing different approaches to vision-action learning.

Building a 6x Faster GraphRAG: Typed Graphs, FAISS, and Two-Phase RetrievalRolodexA custom GraphRAG system optimized for structured data like GitHub repos, achieving ~6x faster indexing with typed entity schemas and two-phase retrieval.

Visualizing Backpropagation in Neural Network Training at Any ScaleTowards Data ScienceUsing HiPlot to generate parallel coordinate plots to visualize deep learning model training.

Building a Real Time Chat Application with NLP CapabilitiesTowards Data ScienceA chat app with sentiment analysis and tone detection using TensorFlow JS Deep Learning API, IBM Cloud, Node.JS, Web Sockets, and React.