Graphsignal

Graphsignal is a production-scale inference profiling platform that helps engineers optimize AI performance across models, engines, GPUs, and other accelerators. I

https://github.com/graphsignal/graphsignal-profiler

TIRx: An Open Compiler Stack for Evolving Frontier ML Kernels

TIRx introduces an open-source, hardware-native DSL and compiler stack for writing high-performance ML kernels that can keep pace with rapidly evolving AI hardware. Built on Apache TVM, it combines low-level control, compiler-assisted abstractions, and agent-friendly tooling to support expert developers, automated kernel generation, and future accelerator architectures.

https://tvm.apache.org/2026/06/22/tirx

Data-Juicer

Data processing for and with foundation models!

https://github.com/datajuicer/data-juicer

Recall

Stop wasting tokens and re-explaining your project every session. Recall gives Claude Code durable memory - entirely offline.

https://github.com/raiyanyahya/recall

Use AI to not use AI (as much)

Rather than using AI to solve the same problem repeatedly, the author suggests using it to build tools that solve the problem for you. A practical example demonstrates how an AI-generated Python script transformed a one-time GitHub usage analysis into a reusable workflow.

https://www.raymondcamden.com/2026/06/22/use-ai-to-not-use-ai-as-much

TimeCapsuleSMB

Hacking the Apple Time Capsule to run modern Samba.

https://github.com/jamesyc/TimeCapsuleSMB

Moebius

Lightweight Image Inpainting Framework with 10B-Level Performance.

https://github.com/hustvl/Moebius

gzipt

A compression based language model.

https://github.com/nathan-barry/gzipt

Build and Deploy a Remote MCP Server to GKE in 30 Minutes

Learn to build and deploy a secure remote MCP server on Google Kubernetes Engine (GKE) in 30 minutes. Streamline your AgenticAI workflows with standard protocols.

https://cloud.google.com/blog/topics/developers-practitioners/build-and-deploy-a-remote-mcp-server-to-gke-in-30-minutes/

Build real agentic apps using CUGA: two dozen working examples on a lightweight harness

Building an agent is mostly plumbing: tools, state, guardrails, scaling from one agent to many. CUGA (pip install cuga), short for Configurable Generalist Agent, the Agent Harness for the Enterprise from IBM handles that, so you write just a tool list and a prompt. We built two-dozen single-file apps to prove it. Read one end to end here, then see how the same agent runs sovereign and go...

https://huggingface.co/blog/ibm-research/cuga-apps

Unlimited-OCR

Unlimited-OCR is Baidu’s open-source OCR model for one-shot long-horizon document parsing, designed to read multi-page documents in a single pass instead of chunking them.

https://github.com/baidu/Unlimited-OCR

Building AI Agents You Can Trust & Evaluate | Agentic RAG Live Series

Hands-on live series building Agentic RAG systems with retrieval, tool calling, evaluations, traces, and production engineering practices.
140 min marathon

https://youtube.com/live/_44x_yL7Q-s?feature=share

OpenKB

Open LLM Knowledge Base.

https://github.com/VectifyAI/OpenKB

Have your agent record video demos of its work with shot-scraper video

shot-scraper video is a new command introduced in today's shot-scraper 1.10 release which ...

https://simonwillison.net/2026/Jun/30/shot-scraper-video/#atom-tag

Scripting good practices in Python

SummaryYou can make your scripts just a little nicer with some tweaks:Use inline dependencies if ...

https://www.bitecode.dev/p/scripting-good-practices-in-python