Improbable: Engineering Worlds That Shouldn't Be Possible

Founded in 2012 by Herman Narula and Rob Whitehead, both graduates of the University of Cambridge, Improbable set out with an ambitious thesis: that the virtual worlds we build are held back not by imagination but by the limits of the computing infrastructure underneath them. The company's London headquarters, nestled in the streets of Spitalfields, has grown from a scrappy startup into one of the UK's most significant deep-tech ventures, having raised over $600 million in funding from backers including SoftBank and Andreessen Horowitz.

The Core Technology

At Improbable's heart is a platform called SpatialOS, a cloud-based operating environment designed to let developers create simulations and virtual worlds of a scale and complexity that a single game server could never support. Traditional multiplayer architectures split a world into shards or instances, capping how many entities can exist and interact in the same space. SpatialOS takes a fundamentally different approach, distributing the simulation across a dynamic mesh of servers so that thousands, even tens of thousands, of concurrent participants and entities can coexist in one coherent reality.

This is not just a gaming technology. While Improbable's early partnerships were with game studios, the same infrastructure has found applications in defence, urban planning, and large-scale event simulation. The company's M&S (Modelling and Simulation) division works with governments and defence organisations to model complex scenarios, from military logistics to disaster response, at a fidelity that was previously impractical.

From Metaverse Ambitions to Pragmatic Focus

Improbable rode the metaverse wave of 2021 and 2022 with visible enthusiasm, launching MSquared (now branded as a distinct initiative) to build an open, interoperable network of virtual worlds. The vision was sweeping: a metaverse not owned by any single corporation, underpinned by decentralised technology and web3 principles. The company invested heavily, acquiring studios and forging partnerships to populate this network with experiences.

As the broader industry's metaverse hype cooled, Improbable recalibrated. The company shed some of its gaming studio acquisitions and refocused on what it does best, building infrastructure. This pragmatic pivot is characteristic of the company's leadership. Narula, still CEO, has spoken publicly about the difference between believing in a long-term vision and being disciplined about the path toward it.

"We are not building a product for next quarter. We are building the computational fabric for a category of experience that doesn't fully exist yet. That requires patience, rigour, and people who are energised by hard problems rather than easy wins."

The Engineering Culture

Improbable's engineering organisation is, by necessity, unusual. The problems the company tackles sit at the intersection of distributed computing, real-time networking, game engine integration, and cloud orchestration. Teams tend to be small and autonomous, organised around specific technical domains rather than product features. A networking team might spend months optimising how state is replicated across server boundaries, while a runtime team works on the scheduler that decides which servers handle which parts of the simulation.

The company uses a mix of languages and tooling. C++ and Rust feature heavily in performance-critical systems. Go and Java appear in platform and infrastructure services. There is a strong internal culture of technical writing, with design documents and post-mortems treated as first-class artefacts. Engineers are expected to articulate not just what they built but why they made the choices they did.

London remains the primary engineering hub, though Improbable maintains teams in other locations. The Spitalfields office reflects the company's personality: open, somewhat understated, with an emphasis on collaborative workspace rather than flashy perks. That said, the compensation and equity packages are competitive with what top-tier London tech companies offer, a reflection of how seriously Improbable takes the calibre of the people it hires.

Working at the Edge

What distinguishes Improbable from many London employers is the nature of its technical challenges. Engineers here are not optimising ad-click pipelines or building yet another SaaS dashboard. They are solving problems in distributed consensus, spatial partitioning, entity lifecycle management at scale, and real-time synchronisation across continents. For the right kind of engineer, this is deeply compelling work.

The company has also invested in its defence and simulation business, which brings its own set of challenges. Security clearances, compliance requirements, and the need for deterministic, reproducible simulations add layers of complexity that consumer-facing products rarely encounter. Engineers who work across both the commercial and defence sides of the business gain an unusually broad perspective.

What Lies Ahead

Improbable's trajectory is not without uncertainty. The metaverse as a cultural moment has faded from headlines, and the company must demonstrate that its technology can sustain viable businesses in the near term while keeping its longer-term bets alive. The defence and simulation work provides steady revenue and credibility. The virtual worlds platform, meanwhile, continues to evolve.

For prospective employees, the question is whether you are drawn to a company that is building something genuinely novel, with all the risk and reward that entails. Improbable is not a safe bet in the way that a large established tech firm might be. It is, however, one of the few places in London where you can work on problems that push the boundaries of what distributed computing can do, in service of experiences that millions of people may one day inhabit.