platoseedplatoseed
Unlock wasted GPU capacity.
Expanse unlocks wasted GPU capacity. We recover idle compute through three capabilities: resource prediction (right-sizing job submissions before they reach the scheduler), optimisation suggestions (code and config changes researchers can apply themselves), and failure prediction (catching jobs that will fail before they consume hours of GPU time). We’re four engineers. We ran HPC and GPU training workloads at the largest quant funds and national supercomputing centres. We faced this problem first hand and the only fix was to over-provision and burn millions. Ismaeel built the first multimodal HPC resource predictor as research at EPCC (Edinburgh’s Parallel Computing Centre), which beat every published baseline. This is the tool we wish we had.
Expanse is an intelligence layer for HPC and GPU clusters that analyzes telemetry from jobs to predict resource fit and failure risk, enabling higher throughput without extra hardware. It aims to unlock unused GPU capacity by advising pilots and providing capacity reports.
Expanse deploys with a seamless one-click setup that streams telemetry passively with no user-side changes. It builds a living knowledge base from cluster history and predicts fit and failure risk at submission time, then provides a capacity report highlighting hidden utilization and how much Expanse could unlock. A pilot can be proposed based on the report, with a multi-week observation period and ongoing support.
Who it’s for: Organizations using HPC or GPU clusters seeking to improve utilization and predictability of workloads; likely labs, hedge funds, and other compute-heavy environments.
backed by Y Combinator; mentions pilot programs and multi-week deployment
Ismaeel is co-founder and CEO of Expanse. Built the first multimodal HPC resource predictor at EPCC (Edinburgh’s Parallel Computing Centre), beating every published baseline. Previously: ran large scale ML models at one of the world’s largest quantitative funds (QRT). Studied Computer Science at the University of Edinburgh.
Niko is co-founder and CTO of Expanse. Trained and optimised speech recognition models on GPU clusters. Previously: managed the platforms researchers and engineers depended on at one of the world’s largest hedge funds (Millennium). Studied Computer Science at the University of Edinburgh.
Yafet is co-founder and COO of Expanse. Built the first GNN-based cluster graph network for predicting SLURM queue wait times at EPCC (Edinburgh’s Parallel Computing Centre). Previously: tooling and infrastructure for researchers at one of the world’s largest quantitative funds (G-Research). Studied Computer Science at the University of Edinburgh.
Eren is co-founder and CPO of Expanse. Built state-of-the-art decentralised foundation model training systems and performance models. Previously: prototyped emerging technologies in quantitative finance (G-Research). Studied Computer Science at the University of Edinburgh.
Unlock wasted GPU capacity.
Expanse optimizes GPU cluster utilization by analyzing code and hardware telemetry to predict optimal resource configurations and detect failures. The platform runs locally on-premise for data sovereignty and reduces debugging time from days to seconds for compute infrastructure teams.
▲ 47
AI for AI Infrastructure
Distributed Compute for AI
