get hired at
Expanse
Predicts what every AI workload needs before it runs, on your own cluster
- 4
- CVPPVAAS+2
- San Francisco
- B2B
Open roles at Expanse
3 positions we're tracking, re-checked daily
- apply ↗
Founding Engineer, ML systems (resource and failure prediction models)
san francisco, ca, usaMid-levelfirst seen 2 weeks ago - apply ↗
Founding Engineer, scheduler integrations (SLURM, Kubernetes, Nomad)
san francisco, ca, usaMid-levelfirst seen 2 weeks ago - apply ↗
Founding Engineer, deployment and observability (Ansible, Helm, self-hosted)
san francisco, ca, usaMid-levelfirst seen 2 weeks ago
Know when Expanse is hiring before anyone else
A role stays uncontested for about four days. Here's the window, and where we put you in it.
From $5.99/week, cancel any time.
- live
- +2h+2hrole spotted & verified1
- +4h+4hyou get the alert1
- day 3day 3you've applied~8
- day 7+day 7+hits the job boards250+
What Expanse is building
Expanse is a prediction layer for GPU and HPC clusters. Every job submitted to SLURM, Kubernetes or Nomad today starts with an engineer guessing how much memory, how many GPUs and how long it needs, and then over-requesting to be safe. Expanse Analyse replaces the guess with a forecast: before the job queues, it predicts memory, runtime and failure risk as confidence-bounded distributions and hands the scheduler a resource request. The scheduler keeps control; Expanse only supplies better numbers. Expanse Diagnose is the other half of the loop. After a job finishes or dies, it reads the logs, metrics and exit state, ties the outcome back to the prediction, and explains the root cause. Each completed job trains the next prediction, so the model becomes specific to your cluster and your workloads over time. Everything is self-hosted: the models run on your GPUs, the management layer on your servers, and no code, logs or telemetry leave the cluster, which is the only way this sells into hedge funds and national labs. The origin is academic and specific. Ismaeel Bashir's master's dissertation at Edinburgh Parallel Computing Centre was 'Multi-modal HPC Resource and Failure Prediction', and the YC profile says it beat all published baselines. Yafet Melake built a GNN over the SLURM cluster graph for queue prediction at the same lab. The company is that research productised, installed with one Ansible playbook and a Helm chart.
Why this matters
The waste numbers are well documented. Microsoft Research has measured GPU utilisation of roughly 50% across many internal deep-learning workloads, and Expanse's own estimate is that around 30% of cloud spend is lost to over-allocation. In one early production deployment Expanse says it identified nearly $8M of idle compute capacity in a single month. Every lab and fund that bought GPUs in the last two years is now under pressure to show they are used, and the schedulers they run on (SLURM especially) have no idea what a job needs until it fails. Crane's Scott Sage put the thesis plainly: the future of AI 'won't be defined only by who builds the biggest clusters, but by who uses them most intelligently.'
Investors: Crane Venture Partners (Scott Sage), PXN Ventures (Andy Barrow), AIP Seed, Y Combinator, Angels: former DeepMind researchers and AI infrastructure leaders
Hiring outlook
The press release states the round will 'expand its engineering team' and Daily Business ran the story under the headline 'ready to expand team', but there is no careers page, no YC job listings, and no named roles as of 17 Sep 2026. Four-person team, so the next hires are founding engineers by default.
Working at Expanse
Expanse is a B2B, Infrastructure company based in San Francisco, CA, USA. Working at a B2B, Infrastructure company at this stage means broad remit, direct access to founders, and equity that still means something if the company works out.
San Francisco, CA, USA is where most Expanse positions are located.
How to actually get hired at Expanse
Why applying the normal way doesn't work
Expanse is <50 people. No recruiter, no HR. Your application goes to a founder juggling everything. Cold outreach is how you actually get seen.
Who to contact at Expanse
- Who decides
- a founder or department head
- Best channel
- LinkedIn or direct email
Founding team
University of Edinburgh integrated master's in Informatics (2021-2026), first-class honours with the class prize; master's dissertation on multi-modal HPC resource and failure prediction at Edinburgh Parallel Computing Centre, which the YC profile says surpassed all published baselines. Previously led large-scale ML models at quant fund QRT. Founded Expanse in October 2025 while still a student.
University of Edinburgh computer science. Trained and optimised speech-recognition models on GPU clusters, then managed the platforms researchers and engineers depended on at Millennium, one of the world's largest hedge funds. Co-founders Eren Mendi (CPO, ex-G-Research, Edinburgh and Imperial) and Yafet Melake (COO, ex-G-Research, built a GNN-based SLURM queue predictor at EPCC) round out the team of four.
What to show them
Take a public HPC job trace (the Alibaba GPU cluster traces or a SLURM accounting dump) and train a model that predicts peak memory and failure with calibrated intervals, then show the over-allocation you would recover at a given confidence level. That is the product, and it is the paper Ismaeel wrote.
A cold email that works at Expanse
What Expanse screens for
Crane leads early infrastructure rounds in Europe and PXN and AIP Seed are Scottish and UK deep-tech seed funds; the angels are ex-DeepMind, which tells you the diligence was on whether the prediction models are real. YC P26 plus a $5.3M seed nine months after founding is a fast track, and the company has already moved to San Francisco to sell. Four engineers and money earmarked for engineering means the next three to five hires define the company. The edge for a candidate is HPC literacy: the target list is quant funds, life-sciences labs and national supercomputing centres, and very few ML-infrastructure engineers have run jobs on a real SLURM cluster.
Customize your CV for the Expanse role. Matching the job description language helps clear ATS filters.
Don't make these mistakes
Ismaeel wrote his dissertation on exactly this problem and beat published baselines before founding the company; do not explain to him that GPU utilisation is low or that SLURM is old. Do not pitch 'AI for FinOps' or cloud cost dashboards; Expanse is self-hosted and lives at the scheduler, not the billing console. Do not lead with Kubernetes-only experience if you have never touched an HPC queue; the first customers run SLURM. Lead with a specific job you have watched OOM at hour 11 of a 12-hour run, and what signal in the logs or code would have predicted it.
Mistakes that kill Expanse applications
A recycled CV gets rejected fast at Expanse (<50 people). They notice.
Don't open with what you want. Open with what Expanse is dealing with right now — The waste numbers are well documented, and what you'd do about it.
Most B2B, Infrastructure applications get ghosted. A day-five follow-up can double your response rate.
Applying to Expanse? Get the contact, not the form.
The Expanse interview process
We don't yet have verified candidate reports for Expanse. What follows is the typical process for a <50-person B2B, Infrastructure company, so treat it as a model, not confirmed detail.
Interview stages
- 1
Intro Call
Video call · 30 min
What it testsCulture fit and role expectationsUsually run byFounder or hiring manager - 2
Technical Deep Dive
Video call or in-person · 60 min
What it testsPast projects and problem-solving approachUsually run byTechnical founder or lead - 3
Final Round
In-person or video · 45 min
What it testsTeam fit and offer discussionUsually run byFounding team
Expanse interview timeline
Expect 7 days total. Compared to similar B2B, Infrastructure companies (10 days median), Expanse is faster.
Interviewed at Expanse?
Tell us how it went: stages, questions, timeline. Takes 90 seconds and it's how this page stays accurate for the next person.
Expanse jobs, frequently asked questions
How many jobs does Expanse have open?
As of October 2, 2026, Expanse has 3 open positions.
Does Expanse hire remotely?
Expanse doesn't have remote openings at the moment. All roles are in San Francisco, CA, USA.
What roles is Expanse hiring for?
Expanse is hiring across Engineering. The most recent opening is Founding Engineer, ML systems (resource and failure prediction models).
How do I apply for a job at Expanse?
Use the apply links above, or check our guide to getting hired at Expanse.
Does Expanse respond to cold emails?
Not enough data yet on Expanse's cold email response rates.
Who is the hiring manager at Expanse?
At this size, hiring is usually run by a founder or department head.
How competitive is it to get hired at Expanse?
<50-person B2B, Infrastructure companies see ~50-100 applicants per role in two weeks. The 72-hour window is your advantage.
How many rounds is the Expanse interview?
3 stages: Intro Call, Technical Deep Dive, Final Round.
Is the Expanse interview hard?
Expect relevant experience and culture fit, not algorithm trivia. Candidates report Final Round as the toughest stage.
Does Expanse give a take-home task?
No, Expanse does not include a take-home stage.
How long does Expanse take to get back to you?
Around 7 days across the full process.
What should I prepare for the Expanse interview?
Focus on relevant experience and culture fit. At <50 people, they're testing whether you can operate without process, not whether you memorised algorithms.
Where is Expanse based?
Expanse is headquartered in San Francisco, CA, USA.
Get Expanse roles before they're posted
A role stays uncontested for about four days. Here's the window, and where we put you in it.
From $5.99/week, cancel any time.
- live
- +2h+2hrole spotted & verified1
- +4h+4hyou get the alert1
- day 3day 3you've applied~8
- day 7+day 7+hits the job boards250+
Related
More B2B, Infrastructure companies hiring
Other startups we track in the same space.