Hacking on new things for fun and learning. I build agent driven systems, data pipelines, and whatever else needs to exist to make world a better place.
- 🔬 SutroYaro – Agent-driven research lab with the Sutro Group [📝 writeup]
- Hinton Problems – 55 papers reimplemented as runnable catalogs [📝 writeup]
- Schmidhuber Problems – 58 papers reimplemented as runnable catalogs [📝 writeup]
- 🧮 Bourbaki – Autonomous agent for mathematical reasoning and proof [📝 writeup]
- 🧭 Erdos Navigator – Agent navigator that turns Claude Code into a theorem proving assistant [📝 writeup]
- 🎓 AgenticEngineer.ING – Interactive lessons to become an agent-driven engineer across any domain
- 🍄 Mycelium Theme – Digital garden, Obsidian workflow, and blog in one [📝 writeup]
- 📡 StackUnderflow – Local observability for your coding agents: search, replay, and analyse sessions offline. Starts with Claude Code
- 🧩 Chimera – Coding agent framework composed from modular primitives [📝 writeup]
- ✍️ yad.codes – Technical blog and digital garden
- 🚀 Novalis ⏳ – AI-native terminal designed to be rewritten by the coding agents running inside it
- 🗄️ MCPantry ⏳ – Desktop app to manage MCP servers, skills, plugins, and agent configs across your AI tools
- 🌐 CodingAgents.Dev ⏳ – Link aggregator and community for coding agents
- 🎮 Claude Code Games 101 ⏳ – Learn Claude Code by building games
- Cofounded KayOS – Built 13 production AI platforms across legal tech, marketing analytics, and BI. Developed agent infrastructure (knowledge graphs, AI sandboxes, ETL frameworks) and a full AI coding IDE with ACP integration. 21k+ commits across 65 repos.
- Before that: Data / ML at NEAR Foundation (770M+ blockchain transactions), Be On Deck, and RobyAI.
- Contributed to Keras and TensorFlow – bug fixes and API improvements in the early deep learning ecosystem
- Desc2Code – Early research on neural code generation from natural language descriptions
- One-Hundred-Layers-Tiramisu – Semantic segmentation in Keras
- Nvidia-Autopilot-Keras – End-to-end self-driving in Keras
- Keras-SegNet-Basic – SegNet for semantic segmentation in Keras
- Self Healing and Improving Agent Systems – LLM long-horizon tasks, vision/world models, SecResearch/DevSec, sandboxes, VMs
- Program Synthesis / Coding Agents – programsynthesis.pub
- RL and Neuroscience – curated 432 papers in Study-Reinforcement-Learning
- Comedy and Creative Writing – built Library of Comedy, the internet's largest standup database
- Autodidacticism and Superlearning
- yadkonrad.com – personal site
- programsynthesis.pub – research
- @yadkonrad – Twitter