Expanse GTM brain

Expanse predicts GPU job resource needs and failures for teams operating HPC and AI clusters.

category

GPU workload optimization

market profile

  • text: Training, inference and simulation workloads require GPU, memory and runtime allocations before execution, often based on estimates.
  • text: GPU scheduling and cluster management platforms overlap with optimization, although Expanse positions its prediction layer alongside existing schedulers.

personas

  • title: Head of AI infrastructure
    pains:
    • Expensive GPUs sit idle while jobs are overprovisioned.
    • Capacity planning is difficult when job requirements are uncertain.
  • title: HPC cluster administrator
    pains:
    • Users request too much memory or runtime to avoid job failures.
    • A new tool must work with existing schedulers and operating practices.
  • title: ML researcher or training engineer
    pains:
    • A long training run can fail after consuming hours of GPU time.
    • Choosing appropriate GPU and memory allocations requires repeated experimentation.

icp

  • text: Organizations operating their own GPU or HPC clusters for AI training, inference or simulation, including research and quantitative finance teams.
  • text: Clusters using SLURM or Kubernetes with recurring workloads and enough expensive capacity for utilization improvement to matter; minimum cluster size is unverified.

message pillars

  • text: Increase useful output from existing hardware.
  • text: Predict resources and risks before execution.
  • text: Integrate into the current cluster instead of replacing its workflow.

objections

  • objection: Our workloads are unique, so generic predictions may fail.
    response: Evaluate prediction error and saved capacity against historical jobs, including confidence intervals, before expanding.
  • objection: We cannot replace our scheduler or expose sensitive workloads.
    response: Review the existing-cluster integration and validate deployment and data handling with security and infrastructure teams.
  • objection: Our monitoring already shows GPU utilization.
    response: Compare retrospective utilization dashboards with pre-run resource and failure predictions on a representative workload.

competitors

  • name: Run:ai
    why: GPU orchestration and resource management competes for the same cluster efficiency budget.
  • name: Exostellar
    why: HPC and AI resource optimization platform identified in a competitive assessment.
  • name: Altair PBS Professional
    why: Established HPC workload management and scheduling alternative.

influencers

  • name: Ismaeel Bashir
    url: https://expanse.sh/company
    why: Co-founder and CEO articulating the compute certainty positioning.

keywords

  • GPU cluster utilization
  • GPU workload prediction
  • SLURM resource optimization
  • HPC job failure prediction
  • AI training GPU right sizing
  • GPU capacity planning

voice

  • Technical and evidence-led
  • Direct about wasted capacity
  • Engineer-oriented

signals

  • text: An organization expanding its GPU fleet while reporting long queues or poor utilization is a candidate for a capacity audit.
  • text: SLURM or Kubernetes cluster operations and frequent training or simulation jobs indicate a relevant integration environment.

stack

  • SLURM
  • Kubernetes
  • GROMACS

pricing

model: unknown

motion

sales_led

channels

  • Website discovery calls
  • Technical documentation
  • Company blog
  • YC directory

confidence

medium

Founders

  • name: Ismaeel Bashir
    title: Founder/CEO
    role: CEO
    bio: 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.
  • name: Nikodem Bieniek
    title: Founder
    role: Founder
    bio: Nikodem 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.
  • name: Yafet Melake
    title: Founder
    role: Founder
    bio: 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.
  • name: Eren Mendi
    title: Founder
    role: Founder
    bio: 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 and Imperial College London.

Back to YC companies