Co-founder & CTO — GPU Compute Exchange

Co-founder & CTO — GPU Compute Exchange Location: Europe (strongly preferred); exceptional candidates in the United States considered Timing: As soon as possible, ideally joining while the seed round is in process Company: (name confidential) The opportunity My Client is building an exchange for idle AI compute. The AI build-out is creating a very large pool of GPU capacity that is paid for but under-used: take-or-pay contracts bill every GPU-hour, while most enterprises running their own GPUs report using half of that capacity or less. The Company pools idle GPUs from neoclouds, data centers, GPU funds and lenders into one cloud and sells the hours on demand. It then adds managed services, sells each GPU many times over as per-token inference outcomes, and builds towards price data and trading on the resulting flow. The hardest part is already built. The Company runs on a production-proven, multi-tenant GPU cloud stack developed over eight years and two full rewrites. The stack has served more than 5,000 B2B accounts, runs AMD, NVIDIA and Intel accelerators in production, and has been licensed to a neocloud that launched at a $2.4bn valuation, passing NVIDIA’s technical diligence along the way. The stack’s IP is being assigned to The Company, which will own all new development. My Client is backed by an established European technology holding that has operated compute infrastructure through several boom-and-bust cycles for more than a decade (over 1 GW operated). The company is raising a seed round to fund the engineering team. We are looking for a technical co-founder to own the platform, turn the existing stack into the exchange and build the engineering team. What you’ll lead - Own the technology vision and product roadmap across the four stages of the business (aggregate, manage, outcomes, trade), with each stage live and earning revenue before the next one starts. - Get aggregation live within 3–6 months of the seed closing: a fast, repeatable way to onboard third-party GPU clusters into one cloud across sites and GPU generations (e.g. H100, H200, B200, MI300), without owning the hardware. - Extend multi-tenancy from within a single cluster to across sites, so that capacity from many suppliers appears to buyers as one cloud behind one API. - Finish and launch managed services on partner hardware: managed Slurm and Kubernetes, storage, support and SLAs. - Build the outcomes layer: serverless inference sold per token, with one scheduler across GPU generations and safe sharing of each GPU among many buyers. This oversubscription model is at the core of The Company’s economics. - Lay the technical foundations for metering, pricing and utilization data that will later support a compute price index and hedging products. - Harden the existing codebase in the first six months, addressing known technical debt, and lead security and compliance, including SOC 2. - Validate the technical differentiation early: identify where suppliers and buyers get measurably better economics or service from Company X than from existing GPU marketplaces and inference providers, and turn that into product decisions. - Work directly with suppliers (neoclouds, data center operators, GPU funds and lenders) and buyers (AI-native companies, enterprises, resellers) to translate their needs into the roadmap. - Hire and lead an engineering team of 10–15 people over the first two years, and set the technical operating model. - Represent The Companys technology in the seed and later funding rounds, and work with the CEO on fundraising, partnerships and company planning. The platform The Company’s stack is a general-purpose, multi-tenant IaaS and PaaS platform written for accelerated compute. It was built for public, self-service cloud use rather than adapted from a private cloud or bare-metal hosting platform. In brief: - VM-centric cloud: API-driven virtual machines with familiar, hyperscaler-style workflows, e.g. an 8×H100 machine running in about 90 seconds and about two minutes from sign-up to a running instance. - Kubernetes-based control plane: adapted to orchestrate virtual machines, with high availability and upgrades that avoid coordinated customer downtime. - Own networking and storage: multi-tenant networking across Ethernet and InfiniBand, with object storage and high-performance shared file storage. - Enterprise controls and billing: organizations, teams, IAM-style permissions, budgets and quotas, and per-resource metering; 99.999% uptime on a 280×H100 customer cluster. Not everything is finished, and finishing it is part of the job: cross-site multi-tenancy, managed Slurm and Kubernetes, and serverless inference all need to be completed, and the codebase needs hardening before it scales. What you’ll bring - Senior technical leadership in a GPU cloud, neocloud, hyperscaler, GPU marketplace or comparable cloud infrastructure provider. - Deep software engineering experience designing and operating large-scale, multi-tenant, customer-facing cloud platforms. - Strong knowledge of virtualized compute and orchestration, including Kubernetes and VM-based platforms, and of networking, storage and security in multi-tenant environments. - Hands-on experience with GPU and accelerator infrastructure across vendors and generations, and a practical understanding of AI training and inference workloads. - Experience with inference serving at scale (e.g. vLLM, SGLang, TensorRT-LLM, batching and scheduling) and with GPU-sharing techniques (e.g. MIG, MPS, time-slicing). - A 0-to-1 track record: comfortable being hands-on in a small team, inheriting and hardening a large existing codebase, and hiring an engineering team from scratch. - Enough understanding of data center operations to work credibly with operators on their constraints (power, networking, hardware configurations), without needing to design facilities yourself. - The ability to explain complex architecture clearly to suppliers, customers, investors and non-technical stakeholders. - A builder’s and owner’s mindset: able to set direction, make practical trade-offs and ship each stage quickly. Nice to have: - Experience building marketplace or multi-supplier platforms: federation across sites or providers, metering, billing and pricing. - Familiarity with confidential computing (Intel TDX, AMD SEV-SNP, NVIDIA Confidential Computing, remote attestation) and with EU data-sovereignty requirements. The team - You will join an experienced founder-CEO who previously built and scaled a venture-backed marketplace (raising $30M across seed, Series A and Series B), and a co-founder from the backing holding who leads strategy, partnerships and fundraising. - The original architect of the stack is available as an advisor, and the backing holding provides operational support, supply relationships and access to its network. Engagement and compensation This is a co-founder role from day one. Before the seed closes, you will help shape the technical plan and join investor pitches and technical diligence. From closing, the role is full-time as Co-founder & CTO. The package includes co-founder equity (4-year vesting with a 1-year cliff) and, once the seed round closes, a market salary funded from the round. Specific terms and timing will be agreed with the candidate.