Dogukan Tuna
Building AI agents for material science and autonomous engineering at Supermatter — early stage, heavy development
AI search and multi-agent infrastructure at Manuel AI — HVAC, energy, marine
I think of AI as the ultimate compressor for the hardest problems — and a genuine superpower for how fast teams can move. It's time to build, from bits to atoms. That's what I spend my time on: multi-agent systems, reinforcement learning infrastructure for LLMs, high-compute RL, megakernels, and large-scale retrieval and memory.
Mem-RLM — Memory-Augmented Inference for Recursive Language Models
LLM inference, memory augmentation, recursive language models, reinforcement learning, AI agents, open source, test-time compute, strategy learning
An open-source memory layer for Recursive Language Models that records execution trajectories, extracts reusable strategies, and injects them into future runs. Models stop starting cold and actually learn which approaches work for which problem types — 26% accuracy improvement on weaker models, fully stateful inference.
Feb 23, 2026
ContextJira — AI-Native Context Extraction from Jira
developer tools, Chrome Extension, Jira, AI workflow, LLM, productivity, open source, AI developer tools
A Chrome Extension that turns any Jira issue into structured Markdown you can paste straight into Claude, ChatGPT, or any LLM. One click, full context — no manual reformatting, no lost details.
Feb 19, 2026
Teaching agents GPU/TPU/QPU compute: An open skill catalog
GPU programming, TPU, QPU, quantum computing, HPC, CUDA, agent skills, open source, accelerated computing
An open-source, growing catalog of structured agent skills for GPU/TPU/QPU-accelerated frameworks and simulators. The goal: cover every mature and experimental HPC compute workload so any agent can pick up working knowledge at inference time.
Feb 15, 2026
Claude Code-Time Skill Acquisition with Agent Teams
AI agents, Claude Code, multi-agent systems, skill acquisition, agent coordination, knowledge base, React Native
A team of agents researched, synthesized, and integrated a production-grade React Native skill into a shared knowledge base in under 15 minutes — just through coordination at Claude Code-time.
Feb 7, 2026
On Compression, Computation and the Space Between
Kolmogorov complexity, information theory, computation theory, neural networks, Wolfram, compression, mathematics
Kolmogorov complexity, neural networks as program search and Wolfram's ruliology seem to be looking at the same thing from different rooms.
Feb 1, 2026
Defeating Nondeterminism in LLM Inference: Reproducing Batch-Invariant Ops (RMSNorm & Tiled Matrix Multiplication) in JAX
JAX, GPU kernels, RMSNorm, matrix multiplication, LLM inference, nondeterminism, batch invariance, deep learning
This learning log is my beginning of a series exploring various kernel-related topics. As a starting point, I will reproduce the implementation of batch-invariant NN operations in JAX, drawing from Thinking Machines Lab's seminal collaborative work, \"Defeating Nondeterminism in LLM Inference.\"
Nov 25, 2025
Streaming deepagents and task delegation with real-time output
LLM agents, streaming, deep agents, multi-agent systems, task delegation, Python, AI engineering
This post demonstrates how to implement streaming capabilities on top of DeepAgents' package with multi-agent setup, with practical code examples and architectural patterns you can apply to your own projects.
Oct 20, 2025
Energetics of Allosteric Communication in Ubiquitin Revealed by Hybrid MCTS-Langevin Simulations
computational biophysics, protein dynamics, Monte Carlo Tree Search, Langevin dynamics, OpenMM, molecular simulation, ubiquitin, allostery
Exploring protein conformational landscapes and identifying potential allosteric communication pathways remain significant challenges in computational biophysics. This study presents a hybrid computational approach combining Monte Carlo Tree Search (MCTS) with Langevin Dynamics (LD) simulations using the OpenMM toolkit to enhance conformational sampling.
May 6, 2025