Head of Partnerships

Bagel Labs is an AI research lab building the default training infrastructure for generative models. Our core technology — Decentralized Diffusion Models (DDM) — decomposes large diffusion training runs into independent specialist sub-models that train across heterogeneous, commodity hardware with zero gradient communication. Resulting in 30–80% reduction in training CapEx with better model quality, peer-reviewed by published academic research.

Our partners and customers are the teams pushing the frontier of video synthesis, world models, and robotics simulation — and spending millions per training run to do it. We make those runs dramatically cheaper.

The Role

We're hiring a Head of Partnerships to own and drive strategic partnerships with robotics companies, AI labs, hardware vendors, and cloud providers — specifically those building or training diffusion-based models for video, world simulation, and robotic control. You'll be the connective tissue between Bagel's research infrastructure and the teams building the physical intelligence stack.

You'll work directly with the CEO to identify, close, and grow partnerships that validate DDM at scale, generate revenue, and establish Bagel as the training stack of choice for the next wave of embodied AI.

What You'll Do

• Build the partner pipeline. Find and prioritize robotics companies, AI startups, and research labs training diffusion models for world models, VLA (vision-language-action) policies, video generation, and robotics simulation. You'll always be hunting.
• Lead technical-commercial conversations. Translate Bagel's DDM architecture and cost-reduction story into something that lands with ML engineering leads, heads of infra, and C-suite decision-makers — and gets them excited to integrate.
• Close design partnerships. Land initial partners who will validate 50–80% training cost reduction on real robotics and world model workloads. These are the deals that prove the thesis.
• Own the robotics vertical. Leverage your network across simulation, embodied AI, manipulation, and autonomous systems to position Bagel as the training backbone for robotics diffusion workloads — the fastest-growing segment in the space.
• Engage hardware and cloud partners. Build relationships with GPU vendors (NVIDIA, AMD, Intel) and cloud providers to bring Bagel's heterogeneous training stack into managed service offerings.
• Feed signal back to R&D. Channel partner feedback on workload requirements, hardware constraints, and model architectures to shape what we build next (e.g., Paris 2.0 video model, VLA model support).

What We're Looking For

• Deep robotics industry network. You know the people building simulation environments, world models, VLA policies, and foundation models for robotics — and they take your calls. Connections at companies like Physical Intelligence, Skild AI, Figure, Covariant, NVIDIA Isaac, Google DeepMind Robotics, Toyota Research Institute, or similar are highly valued.
• Fluency in the diffusion model ecosystem. You understand (at least at a high level) how diffusion models power video, world models, and robotic control — and why training cost is the bottleneck holding everyone back.
• Early-stage partnership or BD experience. You've built partner programs or closed technical partnerships at a seed-to-Series-B deep tech or AI company. You're comfortable with ambiguity and writing the playbook as you go.
• Technical credibility. You can hold your own in a room of ML researchers and infra engineers. You don't need to write CUDA kernels, but you should be able to explain why heterogeneous distributed training matters and why DDM is different.
• Bias to action. Bagel is a ~10-person team that moves fast. We value agency, speed, and taste over process and hierarchy. If you need a 30-slide internal deck before making a call, this isn't the role.
• Bay Area presence. You're based in or willing to relocate to the Bay Area to be close to the densest concentration of our target partners.

Nice to Haves

• Experience in cloud/infra partnerships (AWS, GCP, Azure, CoreWeave, Lambda)
• Familiarity with the GPU secondary market and compute brokerage landscape
• Prior work in sovereign AI, government AI programs, or IRAP-style funding partnerships
• Existing relationships with AI-focused VCs and accelerators (YC, a16z, Lux Capital, etc.)

What We Offer

• Direct access to the CEO and core research team — your input shapes strategy, not just pipeline
• Competitive cash + meaningful equity at a company right at the inflection point
• A front-row seat to one of the most important infrastructure problems in AI
• The chance to build GTM from the ground up at a research-driven company with peer-reviewed, conference-accepted results — and a team that genuinely loves what they're building

#J-18808-Ljbffr