The Surge: 20 Salaried Roles and Counting
A storage company that cut nearly a fifth of its workforce two years ago is now hiring aggressively under leadership poached from Arista Networks and Cisco Systems.
As of May 2025, Qumulo lists 20 salaried roles on Zero G Talent's board, with two added in the past week: enterprise account executives in New York targeting healthcare and media, plus territory managers covering Chicago, the Southwest, and Atlanta. Salary bands for the 20 salaried roles tracked range from $119,000 to $320,000, with a median of $270,000.
| Role Type | Location | Salary Range |
|---|---|---|
| Enterprise Account Executive | New York (remote) | $300k–$350k |
| Territory Account Executive | NYC area | $224k–$320k |
| Strategic Account Executive | New York (sports/media/broadcast) | $250k–$320k |
| Territory Account Manager | Chicago, Southwest, Atlanta | $224k–$320k |
The postings span engineering, sales, and product leadership: a Head of Technical Special Projects in San Francisco, a Staff Embedded Software Engineer focused on embedded AI, and a Software Engineering Manager in Taiwan.
This wave follows a deliberate leadership reset. In July 2024, Douglas Gourlay replaced Bill Richter as president and CEO after an eight-year run. Gourlay brought 70-plus patents and a resume built at Arista Networks, where he ran software as VP and GM, and Cisco Systems. Six weeks later, Michelle Palleschi joined as executive vice president and COO from Sendoso, Skyport Systems, Apple, and Cisco. Their appointments signaled a pivot toward the cloud-native, AI-adjacent workloads that Kiran Bhageshpur, promoted to CTO in May 2022, was hired to pursue.
Qumulo's platform now manages 7 exabytes of capacity across 56 countries for more than 1,100 enterprise customers, Qumulo's figures put. Recent partnerships, including Microsoft for cloud-native storage relief (announced June 2026), Cisco for a CloudBridge architecture that bypasses what Qumulo calls a "400% flash tax," Databricks for NeuralSearch integration, and the launch of NeuralProtect for AI-driven ransomware detection, all require engineering depth in distributed file systems, cloud APIs, and machine-learning caching (NeuralCache claims up to 99% reduction in cloud API costs). The 2022 layoff of 80 employees, roughly 19% of the workforce, was framed as a path to profitability; the current expansion suggests that milestone has been reached.
The company's trajectory (founded in 2012 by Isilon veterans Neal Fachan, Peter Godman, and Aaron Passey, stealth until 2015, $230 million raised by 2018, Wikipedia found, Vancouver office opened in 2019) has always been tied to the unstructured data explosion. The difference now is the buyer: not just media and genomics, but any enterprise running AI workloads across edge, data center, and public cloud. That shift explains why the open roles cluster around distributed-systems engineers who understand consistency models at scale, and sales leaders who can translate "Cloud Data Fabric" into a CIO's budget line.
That technical bar is where the screening starts.
The Technical Filter: What Survives the First Pass
Qumulo's resume screen operates like a distributed-system health check: it looks for specific signals and discards noise. The product — a single multi-protocol file platform spanning on-premises, edge, and public cloud without replication overhead — dictates the technical bar. Recruiters and hiring managers filter for engineers who have built, debugged, or operated distributed storage at scale. Generic "backend" or "cloud" keywords rarely survive the first pass.
The clearest evidence sits in Qumulo's own codebase. As of July 2026, the organization's public GitHub repositories list Python, Shell, PowerShell, HCL, and Rust as top languages. Rust appears alongside Python not as a side experiment but as a primary systems language, signaling that memory safety and performance in the data path matter. Candidates who list Rust with production experience in storage engines, consensus protocols, or high-throughput services move forward.
Distributed file system expertise is the non-negotiable filter. The product's value proposition — a strictly consistent global file system across those environments — means every engineer touches consistency models, metadata scaling, and protocol translation (NFS, SMB, S3, REST). The popular repositories tell the story: qumulo-filesystem-walk, qumulo-api-introduction, cluster-email-alerts, qsplit. These are operational tooling for a clustered file system, not demo apps. A resume that shows work on metadata-heavy workloads, directory traversal at billions of files, or cross-protocol semantics gets a second look.
Cloud-native storage fluency is the second filter. Since Bhageshpur's promotion, Qumulo has launched NeuralCache (an ML-driven caching engine claiming up to 99% cloud API cost reduction) and NeuralProtect (real-time AI-driven ransomware detection), plus an expanded Microsoft collaboration for cloud-native storage. Candidates who have designed tiering policies between NVMe and object storage, implemented S3-compatible APIs, or built control planes that manage fleets across AWS, Azure, and GCP align with the roadmap. HCL (HashiCorp Configuration Language) in the top-languages list signals Infrastructure-as-Code maturity.
System design depth gets tested in the interview loop. The on-site allocates a systems-level design question among three coding interviews. Load balancing and high availability are core concerns for a platform managing 7 EB across 56 countries. Candidates who can articulate trade-offs between consistency and availability during network partitions, or who have tuned consensus implementations for metadata operations, clear this filter.
Python and scripting fluency round out the practical screen. The codebase leans heavily on Python for control-plane services, automation, and test infrastructure. Shell and PowerShell appear for cross-platform operational tooling, as Windows and Linux both run in customer environments.
AI/ML literacy is emerging as a differentiator. NeuralCache and NeuralProtect embed inference into the data path. Engineers who have deployed models at the edge, optimized inference latency, or built feature pipelines for anomaly detection on time-series telemetry stand out. But the bar remains systems-first: the ML serves the storage engine, not the reverse.
The resume screen passes candidates who demonstrate: (1) production distributed file system or object store experience, (2) fluency in Rust or C/C++ for data-path work, (3) cloud storage architecture across at least two major providers, (4) system design rigor on consistency, availability, and partition tolerance, and (5) operational automation in Python and shell. Everything else (Kubernetes certifications, generic microservices, front-end frameworks) is noise unless it directly supports the storage platform.
But technical fluency alone doesn't clear the process.
Culture Fit: How You Operate Under Pressure
Qumulo's support model — "real humans, real answers, no ticket queue" — is cited by customers from Blur Studio's 500 rendering systems to the New Orleans Real-Time Crime Center's 1,000+ cameras. The behavioral screen tests whether candidates frame past incidents around customer impact rather than technical cleverness. The 2022 reduction in force and subsequent leadership reset make retention practical, not theoretical. Candidates who cite problem complexity, customer impact, or technical depth align with a business now expanding its Microsoft collaboration and cloud-native data services.
The underlying philosophy is that skills can be trained but values and attitudes are far harder to change. A bad culture fit drives conflict, lowers morale, and can push out high performers. This logic traces to Qumulo's Isilon heritage: several early employees came from that EMC-acquired storage company, where scale-out file systems demanded tight cross-functional coordination. The behavioral screen operationalizes that heritage: it assumes that distributed systems engineering, like distributed teams, fails silently when communication protocols break down. Candidates who demonstrate they treat interpersonal friction as a debuggable system — observable, instrumented, iterated — pass. Those who treat it as personality chemistry do not.
Those values are pressure-tested in a three-round loop that has held steady.
The Loop: Three Rounds, Four Hours, One Bar
Qumulo's interview process runs three rounds, a structure consistent across multiple candidate reports. The pipeline filters for distributed-systems fluency before a candidate ever reaches an on-site whiteboard, and each stage escalates both technical depth and collaborative signal.
Round 1: Recruiter Phone Screen: 30 Minutes
The opening call is a standard recruiter conversation, but the rubric is specific. Candidates describe a 30-minute discussion that "dived into my experiences and determined if there was a fit between my experiences and what they did at Qumulo." The recruiter maps the résumé to Qumulo's core domains: distributed file systems, cloud storage APIs, and the stack that underpins the platform. This is not a cultural vibe check; it is a competency triage. Candidates who cannot articulate how their past work touches metadata-heavy workloads, scale-out consistency, or POSIX semantics at petabyte scale tend to exit here.
Round 2: Technical Phone Screen: 60 Minutes
The second round is a live coding session hosted on HackerRank with a single engineer on video. The format: a collaborative online IDE, a Qumulo-authored problem, and an interviewer who watches the candidate think aloud. A recent account specifies a tree serialization question (serialize and deserialize a binary tree) but the problem set rotates. Evaluation criteria include clean code, explicit memory ownership reasoning, and the ability to discuss trade-offs between recursive and iterative approaches under latency constraints.
Round 3: On-Site Loop: Four Hours, Same Day
The final round compresses three coding interviews and one behavioral session into a single four-hour block. The coding slate: one graph problem, one implementation-heavy task, and one systems-level design question. The graph question typically involves traversal or shortest-path logic on a topology that mirrors Qumulo's internal metadata graph. The implementation task stresses correctness under concurrency. The systems design prompt is open-ended but grounded in the platform's actual challenges.
The behavioral round runs in parallel, not as a soft-skills afterthought. Interviewers ask for concrete examples of cross-team debugging, production incident ownership, and times the candidate pushed back on a spec to protect data integrity. The signal is binary: does this engineer treat storage correctness as a shared mandate or a ticket to close?
What the Loop Optimizes For
The three-round design filters sequentially: résumé-to-domain match, algorithmic fluency with systems context, then sustained depth across the storage stack. A candidate who clears the phone screen but falters on the graph question usually lacks the metadata-model intuition Qumulo's codebase demands. One who codes cleanly but cannot articulate the design trade-offs reveals a gap between component-level and cluster-level thinking. The behavioral round is the tiebreaker: two equally strong technical performers diverge on whether they have operated in a culture that treats it as a debuggable system and can prove it.
The process is opaque to outsiders; Qumulo does not publish a public interview guide. But the consistency across independent reports suggests a calibrated bar: the company would rather leave a role open than hire an engineer who cannot defend a design decision when a customer's petabyte dataset is at stake.
Candidates who clear the loop share a preparation pattern.
Playbook: What Recent Hires Did Differently
The GitHub repositories Qumulo maintains, including Prometheus and Grafana containers with pre-built dashboards for the Core OpenMetrics API, Python and curl examples for the REST API, and the public documentation portal, are not marketing artifacts. They are the surface area of the platform. Engineers who have cloned those repos, instrumented a test cluster against the OpenMetrics endpoints, or submitted a pull request to the documentation portal arrive at the technical screen already speaking the same dialect as the interviewers.
Open-source contributions carry disproportionate weight when they touch that stack. One who has contributed to Ceph, Rook, or the CSI spec signals familiarity with the same consensus, replication, and failure-domain questions Qumulo's own engineers debate. The company's architecture — a single namespace spanning on-premises, edge, and cloud without replication overhead — means interviewers probe for experience with metadata consistency at exabyte scale, multi-protocol access (SMB, NFS, S3) on the same dataset, and the latency tail of distributed locking.
Storytelling in behavioral rounds follows a different script than at consumer-tech firms. Candidates who adopt that framing rather than technical cleverness align with the value the leadership team emphasizes. The collaboration filter is explicit. The 2022 layoff, followed by the return of founder Aaron Passey and the hiring of Gourlay from Arista and Cisco, reset the cultural contract around low-ego problem solving. Interviewers listen for "we" language that still owns individual contribution over either hero narratives or diffuse credit. The company's 95 Net Promoter Score, unusual for infrastructure software, suggests the bar for cross-functional communication is high: engineers routinely brief sales, support, and field CTOs on architecture decisions.
Cloud-native fluency is now table stakes. The expanded Microsoft collaboration on Azure Native Qumulo, the "1.6 TB/s on AWS" benchmark, and the NeuralProtect ransomware detection feature all sit at the intersection of file data and AI workloads. Candidates who can articulate how a distributed file system serves GPU clusters (checkpointing, feature-store access, model-weight streaming) without replicating data into object stores demonstrate the product intuition the new leadership prioritizes.
For roles touching the edge portfolio (56 countries, deployments on hardware the company doesn't control), candidates highlight experience with heterogeneous environments: ARM and x86, intermittent connectivity, and security postures that forbid outbound telemetry. The Cisco CloudBridge partnership, which bypasses the "400% flash tax," signals that cost-aware architecture is a live product requirement, not a slide-deck talking point.
Coaches converge on one tactical recommendation: treat the recruiter screen as a technical conversation. Qumulo's recruiters are calibrated to spot distributed-systems vocabulary because the engineering team uses those terms in the first phone screen. Candidates who speak the dialect earn the system-design slot; those who default to generic "microservices" language often stall there.
That pattern looks different from what peers demand.
Peer Comparison: Where the Bar Sits Higher
Qumulo's interview filter carries a distinct genetic marker: its founding team — the three Isilon veterans — all cut their teeth at Isilon, the scale-out NAS pioneer that EMC acquired in 2010 for $2.25 billion, TechCrunch reported. That lineage shapes what the company screens for today. Isilon's OneFS file system forced engineers to reason about distributed metadata, locking, and data layout across dozens of nodes, problems that look remarkably like the ones Qumulo now solves across on-premises clusters, edge sites, and AWS, Azure, and Google Cloud simultaneously. When a Qumulo recruiter flags "hands‑on distributed‑storage experience" as a non‑negotiable, they are effectively asking for Isilon‑era fluency: can you debug a split‑brain scenario at 3 a.m.? Can you explain why a global namespace beats replicated silos? That bar is narrower than the generic "storage background" filter used at larger incumbents.
What makes Qumulo's screen unique is the convergence of three vectors that most peers treat separately. First, the product is software‑only: it runs on HPE Apollo, Dell PowerEdge R740xd, and commodity cloud instances alike. Engineers must therefore reason about hardware abstraction layers, NVMe‑oF, and cloud‑provider storage primitives (EBS, Azure Disks, Persistent Disk) in the same breath. Second, the platform targets exabyte‑scale unstructured data (genomics, media rendering, AI training sets) so candidates face system‑design prompts that blend metadata sharding, multi‑protocol (SMB, NFS, S3) consistency, and real‑time analytics via the REST API or qq CLI. Third, the 2022 pivot under CTO Kiran Bhageshpur toward cloud‑native and AI workloads added a new filter layer: experience with Kubernetes operators, CSI drivers, and GPU‑direct storage paths.
What is documented is Qumulo's own trajectory: after a 19% reduction in force in June 2022, the company reopened 20 salaried roles by mid‑2025 under CEO Douglas Gourlay (ex‑Arista, Cisco) and COO Michelle Palleschi (ex‑Apple, Cisco). That rebound, coupled with a $1.2 billion unicorn valuation, Wikipedia's data shows, from the 2020 Series E, signals a hiring bar calibrated for profitability‑era discipline rather than growth‑at‑all‑costs volume. Candidates who clear it tend to carry a specific résumé shape: open‑source contributions to Ceph, Lustre, or Rook; a paper or two on distributed consensus; and a war story about migrating a multi‑petabyte namespace without downtime. That profile is rarer than the typical "five years of storage" checkbox, and deliberately so.
The screen is narrow by design, and deliberately silent on the peripherals.
Out of Scope: Pay, Remote, Funding
This article centers on the mechanics of Qumulo's screening process: the technical filters, cultural criteria, and interview-loop architecture that determine which candidates advance across the company's 20 salaried roles. By design, it does not analyze compensation bands, remote-work frameworks, or financing history. Those topics merit their own treatment; folding them in here would dilute the through-line that matters most to applicants right now: how to clear the specific hurdles Qumulo has built.
Compensation data exists and is visible on the Zero G Talent board. Qumulo's listed roles show a salary band typically ranging from $119k to $320k with a median of $270k across 20 salaried positions, and individual postings such as an Enterprise Account Executive in New York (remote) at $300k–$350k, a Territory Account Executive in the NYC area at $224k–$320k, and a Strategic Account Executive in New York focused on sports, media, or broadcast at $250k–$320k. Territory Account Manager roles in Chicago, the Southwest, and Atlanta similarly post ranges between $224k and $320k. These figures reflect what the platform captures at a point in time; they are not the subject of this piece. Candidates negotiating offers should treat them as reference points, not guarantees, and verify current bands directly with recruiters.
Remote-work arrangements also appear in the board data (several of the same roles are tagged "remote" or "hybrid"), but the article does not map Qumulo's internal policy, eligibility thresholds, geographic constraints, or equipment stipends. Other organizations publish detailed remote-work frameworks covering approval chains, relocation rules, travel reimbursement, and annual agreement renewals; Qumulo's approach may share structural similarities or diverge entirely. That comparison falls outside the screening-process lens. Applicants should clarify location expectations during the recruiter call rather than assume parity with public policies documented elsewhere.
Funding rounds, valuation shifts, and investor timelines are likewise excluded. The company's capital history influences headcount capacity and strategic priorities, but it does not change the questions asked in a system-design review or the behavioral rubric applied in a leadership interview. Candidates who anchor their preparation to financing news rather than the documented technical bar (distributed file systems fluency, C/C++ or Rust depth, cloud storage architecture) misallocate limited prep time.
The scope boundary is deliberate. Every prior section of this article builds toward a single actionable outcome: helping a candidate understand what Qumulo's interviewers are actually evaluating at each stage. Salary, remote policy, and funding details are real and consequential; they are also orthogonal to that outcome. Readers who need those specifics should consult the board listings, the recruiter, and the offer packet, not this analysis.
Twenty roles sit open. The bar is distributed-systems fluency, low-ego collaboration, and a resume that speaks the dialect of the platform. Candidates who clear it won't just pass a screen. They'll recognize the problems waiting on the other side.
Working in frontier tech? Zero G Talent tracks the openings: see every open Qumulo role, browse frontier tech jobs, openings at Overview, and the people building the field.