Engineering VP Josh Clemm on how we use knowledge graphs, MCP, and DSPy in Dash

Dropbox’s VP of Engineering explains how the company built Dropbox Dash, an AI-driven cross-app search and knowledge system that uses indexing, knowledge graphs, and contextual reasoning to unify work content from many tools into a single, context-aware platform. The piece highlights engineering decisions about when to use indexed retrieval over federated approaches, how MCP tool calls a...

https://dropbox.tech/machine-learning/vp-josh-clemm-knowledge-graphs-mcp-and-dspy-dash

bzfs

bzfs is a reliable near real-time, parallel replication and backup command-line tool for ZFS.

https://github.com/whoschek/bzfs

x-algorithm

Algorithm powering the For You feed on X.

https://github.com/xai-org/x-algorithm

Polymcp

Polymcp provides a simple and efficient way to interact with MCP servers using custom agents.

https://github.com/poly-mcp/Polymcp

TTT-Discover

Learning to Discover at Test Time.

https://github.com/test-time-training/discover

Webrockets

High-performance rust powered websocket server for Python.

https://github.com/ploMP4/webrockets

This Design Pattern Scares Me To Death

Scattered business rules and duplicated conditionals slowly rot a codebase by drifting out of sync and creating subtle bugs. The video shows how refactoring with the Specification Pattern makes rules composable, testable, and even configurable as data instead of hardcoded logic.

https://www.youtube.com/watch?v=KqfMiuL3cx4

DynamoDB crash course: part 1 – philosophy

This is part one of a series covering core DynamoDB concepts and patterns, all the way up to single-table design; the goal is to get you to understand idiomatic usage and trade-offs in under an hour. Today, we're looking at what DynamoDB is and why it is that way.

https://death.andgravity.com/dynamodb

mini-swe-agent

The 100 line AI agent that solves GitHub issues or helps you in your command line. Radically simple, no huge configs, no giant monorepo—but scores >74% on SWE-bench verified!

https://github.com/SWE-agent/mini-swe-agent

Creating Reddit Summaries with URL Context and Gemini

Raymond Camden demonstrates using the Gemini API's "URL Context" tool to scrape and summarize Reddit threads without relying on Reddit's restricted APIs. He also shares a "double-prompting" workaround to convert the unstructured URL summaries into a precise JSON format, since Gemini currently disables structured output when using the URL tool.

https://www.raymondcamden.com/2026/02/09/creating-reddit-summaries-with-url-context-and-gemini

distil-text2sql

Query your data in plain English with a fine-tuned Text2SQL model.

https://github.com/distil-labs/distil-text2sql

ACE-Step-1.5

The most powerful local music generation model that outperforms most commercial alternatives

https://github.com/ace-step/ACE-Step-1.5

knowledge-work-plugins

Open source repository of plugins primarily intended for knowledge workers to use in Claude Cowork.

https://github.com/anthropics/knowledge-work-plugins

How the Self-Driving Tech Stack Works

The article breaks down the core components of an autonomous driving stack, including sensing (cameras, lidar, radar), perception (object detection and tracking), planning (behavior and trajectory), and control (actuation and safety). It explains how these layers interact in real time, the trade-offs between different sensors and algorithms, and practical considerations for building and ...

https://cardog.app/blog/autonomous-driving-stack-technical-guide

PyNote

A lightweight, browser-based Python notebook editor that runs entirely client-side using WebAssembly (via Pyodide), so it needs no server or installation.

https://github.com/bouzidanas/pynote-notebook-editor

copier-astral

Fast Python project template with Astral's toolchain (uv, ruff, ty) + pytest, MkDocs, Typer, GitHub Actions, Docker

https://github.com/ritwiktiwari/copier-astral