Member of Technical Staff, Enterprise Foundations

THE ROLE

Enterprise Foundations builds the capabilities the largest companies in the world need before they can run their business on Fireworks. Global banks, insurers, healthcare and legal technology companies, and large-scale SaaS platforms are moving AI into production now, and they arrive with requirements that only apply to scaled companies: how their organization and users are modeled, who is allowed to do what, how usage is metered and billed, what gets logged, where data lives and who holds the keys, and whether the whole thing can run inside their own cloud.

You will own that work as end-to-end features. A single project routinely starts in a customer conversation, becomes a change to a core data model or API, and then has to be threaded through the control plane, the training and inference stacks, the SDK and CLI, and the console before it counts as shipped. Much of what we build is a foundational shift rather than an addition: a new primitive that other teams' code has to move onto, rolled out against live production traffic without breaking anyone.

The work sits close to revenue. Requirements come out of live enterprise deals, security reviews, and customer conversations. You will be in those rooms, decide what product becomes, and go build it.

WHAT YOU'LL DO

• Own foundational capabilities end to end — data model and API through control plane, runtime, SDK/CLI, and console — and migrate live systems onto them

• Design the primitives large customers organize around — organizations and sub-accounts, groups, roles, service identities — and connect them to enterprise directories through federation and provisioning (SSO, SCIM)

• Design multi-tenant authorization, policy enforcement, and audit systems that hold up in front of a security architect

• Make usage metering, spend controls, and billing correct and legible at scale — customers reconcile our numbers against their own

• Own data isolation and key management — customer-managed encryption keys across AWS, GCP, and Azure — threaded through datasets, fine-tuning, RL, checkpoints, models, and inference

• Deploy and operate Fireworks inside a customer's own cloud, and solve data residency and regional isolation for customers with hard constraints

• Use AI tooling aggressively — we expect you to automate yourself

YOU MIGHT BE A FIT IF

• You want your work measured in unblocked revenue, not tickets closed

• You're comfortable owning domains you've never worked in before

• You want the whole problem — data model through UI — and don't need someone else to draw the edges for you

• You can hold your own with an enterprise security architect

• You'd rather ship what closes the deal than perfect the abstraction

• You treat ambiguity as the job, not a complaint about the job

MINIMUM QUALIFICATIONS

• 4+ years building and operating production backend or distributed systems at scale

• Strong server-side engineering skills in Go, Python, C++, TypeScript, or similar, and willingness to work outside your primary layer to land a feature

• Experience designing APIs and data models that other teams build on, and evolving them under live traffic without breaking callers

• Hands-on depth in at least one major cloud (AWS, GCP, or Azure): IAM and workload identity, KMS, object storage, and VPC networking

• A track record of owning an ambiguous, cross-team workstream from requirement through production

• Willingness to work directly with customers

PREFERRED QUALIFICATIONS

• Enterprise identity and tenancy: SSO, SCIM provisioning, RBAC/ABAC, audit logging, and org modeling

• Metering, quotas, or billing systems where correctness is customer-visible

• Applied cryptography and key management: envelope encryption, KMS, and threat modeling

• Kubernetes depth and infrastructure-as-code fluency (Terraform, GitOps)

• Familiarity with LLM inference and fine-tuning — our work runs straight through the training and serving stacks, so it helps

Why Fireworks?

• Solve Hard Problems: Tackle challenges at the forefront of AI infrastructure, from low-latency inference to scalable model serving.