Careers at Helsing: Teams, Pay and How to Get Hired

By August 2026 Helsing carried 129 open roles across 23 departments, concentrated in Munich, Berlin, London, Paris, and a growing Washington, DC presence. This guide explains who the company hires and onto which teams, what the compensation package looks like, how the hiring process works and what distinguishes successful applicants, where the work is performed across its sites, and which personal traits enable employees to thrive there.

The question isn't whether Helsing can attract talent. It's what kind survives the filter.

Helsing calls itself a software company built to keep democracies from harm. That mission draws engineers and researchers who want their work to matter in defence, who argue openly about the ethics of AI in warfare, and who operate at "Helsing speed" — creative urgency sustained by relentless recruitment pressure. The hiring philosophy centers on talent density. Founders decide within 30 minutes of an interview whether a candidate shows the "effortlessness" and "spiky" excellence they prize — unwavering imbalance in at least one domain rather than well‑rounded competence. Leaders calibrate ratings quarterly against a shared bar. Difficult conversations are expected. The result: a workforce skewed toward senior individual contributors who ship autonomously in small, cross‑functional groups.

Who Helsing Hires and the Teams They Join

Engineering and research cores

The technical backbone splits across several clusters. AI research roles — including PhD internships in 3D computer vision posted across Barcelona, Berlin, London, Munich, and Paris — feed directly into products like Cirra (electronic warfare), Centaur and CA‑1 Europa (autonomous air dominance), and the HX‑2 strike drone. Software engineers build the Altra recce‑strike stack and the autonomy layer for SG‑1/Lura underwater systems. Hardware electronics engineers and camera engineers sit in Berlin, London, and Munich, working beside radio engineers and multimodal sensing fusion specialists who integrate sensors onto air, land, and sea platforms. A lead engineer role for ground control stations and hardware simulators signals investment in operator‑in‑the‑loop testing infrastructure.

Programme and product ownership

Customer‑facing programme managers own delivery by domain: Land, Air, and Space. The Space programme manager role based in Munich‑Berlin reflects the push beyond atmospheric systems. Product managers translate mission needs into roadmaps for empowered product teams — Helsing's model for zero‑to‑one innovation, distinct from agile iteration on established products. These roles demand fluency in defence procurement, regulatory constraints, and the technical architecture of the systems they shepherd.

Production, quality, and security

Manufacturing engineers, incoming inspectors, quality inspectors, and a quality process manager are based in Munich. Security engineers and a security manager (recently posted in Paris) protect IP and meet classified programme requirements. Operational buyers, financial accountants, and legal counsel round out the enablement functions that keep a cleared, multi‑national defence company compliant.

Compensation and Equity

Helsing pays to win talent density. As founder Torsten Reil said in a 20VC interview: "We'd rather have a team of four exceptional people than eight good ones. It makes everything easier. It's also cheaper, even though you pay above market rate." That philosophy (backed by a $1.8 billion Series E closed in July 2026) shapes every component of the offer letter.

Salary ranges by role and geography

Third-party aggregators show a wide spread, reflecting both role diversity and the pan-European footprint. Levels.fyi reports total compensation from £39,127 for an Administrative Assistant in Germany to £187,662 for a Program Manager in the United Kingdom. Salary.run's 2026 verified data puts the overall band at €94,000–€232,000 across base, equity, and bonus.

Engineering and AI research roles (the bulk of Helsing's headcount) cluster toward the upper half of that band. The company's own job board lists openings such as Programme Manager – Space (Munich/Berlin), AI Research Intern (PhD) – 3D Computer Vision (multiple sites), Security Engineer (Berlin/London/Munich), and Multimodal Sensing and Fusion Engineer (Berlin/London/Munich), all calibrated to the "above market" benchmark Reil described.

Equity grants and vesting

Helsing uses restricted stock units (RSUs) as the primary equity vehicle. The standard schedule vests over four years with a one-year cliff: 25% after the first anniversary, then monthly or quarterly thereafter. Refresh grants (issued after performance reviews or promotion) sometimes follow a shorter cadence. Privatepulse.online notes that the exact schedule lives in each grant agreement, so candidates should review their specific terms rather than assume a universal template.

Because Helsing remains private, RSUs carry no public-market liquidity. Their value is tied to periodic tender offers or a future IPO — both plausible given the Series E valuation and the company's trajectory into programs of record across European defence ministries. The 2026 fundraise, backed by existing investors including Prima Materia, reset the internal valuation and expanded the equity pool for new hires and refreshes.

Bonus structure

Public data on bonus mechanics is thin. Neither levels.fyi nor salary.run breaks out the variable component separately. Helsing's quarterly calibration process — where leaders review every performance rating against a shared standard of "what good looks like" — feeds directly into compensation decisions.

The talent-density premium

The compensation model is inseparable from the hiring model. Reil's team "hustle like crazy whether it's on LinkedIn whether it's through Network and we mean we won't give up until we get the person that we want." That effort is funded by a cost structure that accepts higher per-head spend to keep teams small. A four-person exceptional team paid at the 90th percentile still costs less than an eight-person "good" team at the 60th percentile — and, in Helsing's view, ships faster with fewer coordination failures. For candidates, the implication is clear: the offer reflects a bet on your leverage, not just your labor.

Helsing's Hiring Process

Helsing runs a six-stage loop that typically spans three to five weeks. The sequence is consistent: a 30-minute recruiter screen, two 45-to-60-minute coding rounds, a 45-to-60-minute system-design interview, a 30-to-45-minute "bring something awesome" session, and a final 30-to-45-minute leadership conversation. InterviewQuery's guide and Glassdoor's aggregated candidate reports both describe this structure, though candidates note communication after the loop can be inconsistent.

The recruiter screen is not a box-ticking exercise. The recruiter walks the candidate through the entire loop, shares preparation guidance, and evaluates whether the person's background aligns with the emphasis on talent density — the principle that four exceptional people beat eight good ones. That philosophy shapes every subsequent round.

The first coding interview focuses on a practical algorithmic problem. Interviewers expect candidates to explain their approach clearly as they work. The second coding round shifts toward LeetCode-medium difficulty with a domain twist; one reported question asked candidates to read a file and trace dependent parts of an aircraft, adding a dynamic-programming flavor. The technical bar is solid but not extreme. What distinguishes successful applicants is the ability to stay grounded when the conversation moves from abstract data structures into infrastructure expectations. Candidates report late-stage Docker requirements that had never been surfaced earlier — a signal that Helsing may carry implicit stack assumptions candidates should clarify early.

System design is where the filter tightens. The prompt is explicitly domain-specific: drones. A documented example asked candidates to design a system that tracks and analyzes crops in the field, including mold detection. This is not a generic architecture exercise. It rewards engineers who can translate messy operational requirements (sensor noise, intermittent connectivity, edge-case failure modes) into concrete components, data flows, and fallback strategies. InterviewQuery frames this as Helsing's real filter: "systems fit," meaning the ability to connect software decisions to a real-world defense or robotics context rather than reciting textbook patterns.

The "bring something awesome" round is deliberately open-ended. Candidates present a project, a technical deep-dive, or a problem they solved that demonstrates spiky excellence. Founders describe the trait they hunt for as "effortlessness" — not trying too hard, but being so on top of the detail that mastery shows without performance. They also look for "unwavering imbalance": a single area where the candidate went all-in, and the mindset that drove it. This round tests whether a person can articulate the why behind their hardest work, not just the what.

The leadership conversation closes the loop. It covers fit, expectations, and sometimes infrastructure experience. Founders say they often know within 30 minutes whether someone is exceptional. They also run quarterly calibration sessions where this calibration process occurs, reinforcing the same talent-density bar used in hiring.

Candidates who clear the loop tend to share three habits: they prepare for LeetCode-medium coding with a practical twist, they study drone-centric system-design scenarios (crop analysis, tracking, sensor fusion), and they ask explicitly about infrastructure assumptions (Docker, CI/CD, deployment environments) before the final round. The process rewards proactive clarification as much as technical depth.

Where the Work Happens

Helsing's physical footprint reflects its evolution from pure-play AI software company into a vertically integrated defense manufacturer. The company operates across a two-tier geography: software and systems engineering hubs in major European cities, and a growing network of "Resilience Factories" built for mass production of autonomous systems. The split is deliberate — the offices write the code and design the architectures; the factories turn them into hardware at wartime scale.

Munich: Headquarters and systems integration core

Munich remains the center of gravity. Founded there in March 2021 by Torsten Reil, Gundbert Scherf, and Niklas Köhler, the Munich office houses leadership, core AI research, and the systems integration teams that stitch software onto partner platforms — Rheinmetall vehicles, Saab sensors, and now Grob Aircraft airframes. A 2023 WIRED profile noted 220 employees spread across four offices, with Munich the largest. Current job postings list Munich for roles spanning Programme Manager – Space, Security Engineer, Camera Engineer, and the aforementioned engineer, signaling active work on sensor payloads, onboard compute, and secure communications. Helsing maintains secrecy around the locations of some of its production sites, remaining wary of sabotage linked to Russia.

Berlin: Autonomy, robotics, and frontier research

Berlin has emerged as the autonomy and robotics hub. The capital hosts the aforementioned internship role and the aforementioned posting, both listed across multiple cities but with Berlin as a primary option. This aligns with Helsing's published work on the Centaur AI agent (an autonomous fighter-jet pilot capable of swarm operation) and the HX-2's onboard navigation stack. A partnership with ARX Robotics, announced September 2024, to develop "AI-based, networked reconnaissance and strike systems integrating drones and unmanned ground vehicles" centers in Berlin.

London: Maritime, space, and UK programme delivery

London serves as the UK sovereign capability hub and the front door for British MoD programmes. The Trinity House Agreement, signed between Germany and the UK, commits Helsing to manufacturing drones at a newly established facility in Plymouth. London teams own programme management, systems engineering, and customer-facing integration for UK-specific variants. Security Engineer postings (Berlin; London; Munich) hint at a distributed security operations center with London as a node. Helsing appointed Andriy Shevchenko as Managing Director for Ukraine in July 2026, coordinating the 4,000-unit HX-2 contract and the HF-1 supply chain with Ukrainian manufacturers including Terminal Autonomy.

Paris: European programme integration and sensor payloads

France's defence procurement agency (DGA) and the Future Combat Air System (FCAS) programme both demand onshore integration capability. Helsing's Saab partnership (September 2023) for integrating AI into existing weapons systems likely has a Paris workstream given Saab's French subsidiary and the FCAS connection. The office also anchors the Estonia subsidiary relationship — Tallinn hosts a Baltic security focus.

Barcelona: Specialized sensing and compute

Barcelona appears only in the aforementioned posting (Barcelona; Berlin; London; Munich; Paris). This is likely a satellite research node — not a full engineering office. The role's emphasis on 3D computer vision aligns with Helsing's push to reduce sensor payload weight and power draw on the HX-2 and future Grob Aircraft platforms.

Southern Germany Resilience Factory: The first mass-production line

The clearest picture of physical infrastructure comes from the first Resilience Factory in southern Germany. According to the company, its first Resilience Factory in Southern Germany is now operational, with an initial monthly production capacity of more than 1,000 HX-2. The facility is designed for surge: "Like other similar facilities, it can also 'scale manufacturing rates to tens of thousands of units in case of a conflict.'" The HX-2's combat deployment in Ukraine — thousands of systems successfully deployed to the front lines per Helsing's July 2026 press release — validates the line's output quality. The factory's exact location is undisclosed; Helsing maintains this secrecy.

West Virginia: The first U.S. Resilience Factory

Announced July 14, 2026, the West Virginia facility was designed as part of Helsing's global network of Resilience Factories and would produce the HX-2, an advanced, AI-enabled strike drone. The company states: "Once fully operational, the West Virginia manufacturing base will be capable of producing more than 2,000 HX-2 drones per month and have the ability to scale further according to demand." The site selection signals intent to serve U.S. DoD and allied programmes directly. The facility will replicate the southern Germany line: final assembly, integration of the Centaur AI compute module, sensor payload mating, and acceptance testing.

Plymouth: UK sovereign manufacturing under Trinity House

The Plymouth facility, announced as part of this agreement, is newly established. No capacity figures are public. Its purpose is sovereign UK production — ensuring HX-2 and follow-on systems can be built on British soil if supply chains are severed. The agreement also implies technology transfer: UK personnel trained on Helsing's production processes, creating a domestic workforce for future programmes like the CA-1 (first flight targeted 2027, entry into service 2029).

Acquired manufacturing capability: Grob Aircraft and Blue Ocean

Two acquisitions extend Helsing's physical plant without greenfield construction. Grob Aircraft, acquired June 2025 from H3 Aerospace, brings a certified light-aircraft production line in Mindelheim, Bavaria. Helsing aims to integrate its AI and software with Grob's manufacturing to develop AI-powered reconnaissance and combat aircraft. This is the CA-1 pathway: Grob builds the airframe; Helsing supplies the autonomy stack and mission systems. Blue Ocean, acquired October 2024, adds Australian autonomous underwater vehicle (AUV) manufacturing. The company seeks to integrate the company's hardware and manufacturing capabilities with its AI technology, opening a maritime Resilience Factory track.

What remains opaque

Helsing does not publish facility tours, floor plans, or equipment lists. The research digest contains zero detail on machine shops, additive manufacturing cells, or electronics prototyping labs at any office location. The Resilience Factory disclosures are capacity figures, not process descriptions. This opacity is intentional — the sabotage concern is real, and the competitive advantage lies in production velocity, not facility showcase. For a candidate, the takeaway is structural: software engineers in Munich, Berlin, London, or Paris ship code to factories they may never visit; manufacturing engineers in southern Germany, West Virginia, Plymouth, Mindelheim, or Australia build the hardware that code runs on. The seam between those worlds — the integration floors, the test cells where Centaur meets airframe, the secure CI/CD pipelines that push updates to deployed swarms — is where Helsing's differentiating work happens.

Traits and Mindsets That Thrive

The people who last at Helsing share a conviction that democratic security is not an abstraction — it is a systems problem with real deadlines. The company's own careers page states it plainly: "We are looking for people with their heart in the right place, who share our conviction that democratic values are worth protecting, for ourselves and for future generations." That sentence functions as a filter. Candidates who treat defence technology as a purely technical optimisation exercise tend to self-select out; the ones who stay treat the mission as a constraint that sharpens every engineering choice.

Mission alignment is necessary but not sufficient. The work sits at the intersection of modern AI, embedded systems, sensor fusion, and military doctrine — domains that historically lived in separate organisations. Helsing's "digital assembly hall" concept, described in a 2024 Defense News report on the European next-generation combat jet programme, captures the reality: AI models have dependencies on one another and on the data they ingest, and someone has to put it all together. Engineers who thrive here are comfortable crossing those boundaries daily. A camera engineer in Munich or Berlin is not just tuning optics; they are reasoning about how their sensor feeds a multimodal fusion pipeline that a colleague in London or Paris will integrate into an autonomy stack that must survive electronic warfare on a Ukrainian front line. The first-party board listings — Camera Engineer, the aforementioned engineer, Security Engineer, the aforementioned internship — map directly to this interdisciplinary surface area.

Operational tempo reinforces the profile. The HX-2 strike drone is already combat-proven in Ukraine, with thousands of systems deployed to front lines and extensive testing by allied forces including the U.S. Army's Flytrap exercise in Lithuania. In early 2025, while 4,000 Germany-funded HF-1 strike drones were being delivered, Helsing agreed to supply a further 6,000 HX-2 units. A West Virginia resilience factory announced in July 2026 targets that rate at full rate. This is not a research lab that occasionally ships demos; it is a production organisation where software updates meet hardware lead times and export controls simultaneously. People who need perfect requirements before writing code struggle. The ones who succeed treat ambiguity as the default state and iterate toward clarity with hardware in the loop.

The scrutiny Helsing faces sharpens the cultural signal. Bloomberg reported in April 2025 that former employees, investors, and military experts raised concerns about overpriced drones, glitchy software, and business-practice integrity. The Rheinmetall partnership fell through in 2024, with the German prime contractor choosing competitor Auterion instead. NGOs and autonomous-weapons experts have warned about machine-vision reliability, opaque decision-making, and the risk that a tactical autonomy mindset gets applied to strategic questions — a concern Defense News analysts underscored when they noted that switching a weapon system from human-in-the-loop to fully autonomous may only require a software change with strong incentives to flip it. Working under that microscope demands a specific kind of intellectual honesty: the willingness to surface failure modes before a customer or a journalist does, and the discipline to document the mitigation so it survives audit. Engineers who treat verification as paperwork rather than a design input do not last.

The Area 9 incubator, launched in January 2025 for "moonshot" projects, signals that the company still bets on long-horizon technical risk alongside the production grind. That duality — delivering thousands of drones this quarter while funding research that may not pay off for years — selects for people who can context-switch without losing rigor. A security manager in Paris or a programme manager for space in Munich and Berlin must navigate classification regimes, supply-chain fragility, and the same AI-model dependencies that the research interns in Barcelona are trying to untangle. The common thread is systems thinking: understanding how a change in training data propagates through a fusion pipeline to a flight controller on a drone manufactured in West Virginia or Plymouth under this agreement.

Founders' backgrounds set the tone. Torsten Reil, a computational biologist from Oxford who founded NaturalMotion (acquired by Zynga), Gundbert Scherf, formerly of the German Ministry of Defence, and Niklas Köhler, a machine learning engineer, built the company from a 2014 conviction after Russia's annexation of Crimea. The pledge to sell only to democratic governments is not marketing copy — it is a strategic constraint that shapes hiring. Candidates who have worked in dual-use or defence-adjacent roles and can articulate why that constraint matters tend to integrate faster than those who discover it at the offer stage.

In practice, the thriving employee at Helsing is someone who has shipped safety-critical or regulated software, understands that "AI-enabled defence" means the model is only as good as the sensor calibration and the supply chain behind it, and treats the mission as an engineering requirement — not a slogan. They write tests for the edge cases that appear when a drone loses GPS over contested territory. They document the assumption that a software switch could change a weapon's autonomy level, and they build the guardrails so that switch cannot be flipped by accident or pressure. They argue about architecture in the open because the cost of silence is measured in lives and credibility. And they come back the next day because the problem set (scaling resilient autonomy for democratic defence) is the hardest one they have found.

Working in frontier tech? Zero G Talent tracks the openings: see every open Helsing role, browse frontier tech jobs, the companies hiring, and the people building the field.