Poster: Real-time multi-hop reasoning with Nvidia cuGraphs
Sep 15, 2025
Presenting research on real-time multi-hop reasoning with Nvidia cuGraphs at PyTorch Conference 2025 (September, California, USA).
The work demonstrates optimized graph traversal techniques for complex reasoning tasks using GPU acceleration, significantly improving inference speed for multi-hop reasoning in large language models.
Keep reading
Why boring technology keeps winningThe most reliable systems I have worked on were built from the dullest parts. That is not an accident.Jul 18, 2026What I learned building Raft from scratchConsensus looks simple in the paper. Then you try to implement leader election with real clocks and real networks.May 2, 2026Give every request a latency budgetp99 problems are rarely one slow thing. They are ten fine things with no budget.Feb 11, 2026AI session on trans-tokenizationConducted a remote session on trans-tokenization for large language models.Mar 15, 2025