Technical Program Manager, Cloud Inference

Anthropic · San Francisco, CA | New York City, NY · Technical Program Management · listed April 28, 2026

The shape of it

Seniority
Manager
Where
Hybrid
Stated pay
$290,000 – $435,000 USD
Requirements listed
6
Length
1,089 words

In the posting’s own words

We are seeking an experienced Technical Program Manager to support our critical cloud deployments. In this role you will be an execution owner, driving coordination and collaboration across multiple engineering teams. You will also support the collaboration and technical execution between our internal engineering teams and our major cloud partners including Amazon Bedrock, Google Vertex, and Microsoft Foundry.

What it asks for · 6

  • Have several years of experience in technical program management, with a track record of successfully delivering complex technical programs, preferably involving cloud platforms and AI technologies.
  • Have strong understanding of cloud computing architectures, AI/ML deployment, and integration challenges.
  • Have exceptional interpersonal and communication skills, enabling you to influence without authority and build cross-organizational support.
  • Have a high threshold for navigating ambiguity and ability to balance strategic priorities with rapid, high-quality execution.
  • Thrive in fast-paced, scaling environments with the ability to bring order to chaos.
  • Are passionate about Anthropic's mission and committed to ensuring AI is developed safely.

Also a plus

  • Direct experience with a hyperscaler's managed AI platform — Amazon Bedrock, Google Vertex AI, or Azure AI Foundry — including how partners list, launch, and onboard customers on it.
  • Background in ML inference, model serving infrastructure, or accelerator-based compute.
  • Have owned joint engineering roadmap or dependency tracking, driving incident follow-through, and converting open issues into a prioritized plan both sides commit to
  • Experience with release engineering, deployment automation, or CI/CD for systems that ship to multiple targets or environments.

What the job covers

  • Partner with engineering leaders to define, scope, and sequence major technical initiatives for cloud partnerships and AI model deployment, and own the plans, timelines, and resourcing to land them.
  • Own launch readiness for Claude models on partner cloud platforms: checklist, blocker tracking, joint go/no-go with the partner, and post-launch stability follow-through.
  • Act as the primary technical interface to cloud partner engineering orgs — owning the relationship, the shared roadmap, and day-to-day coordination on deployment, capacity, and incidents.
  • Drive cross-functional alignment across internal engineering, product, and go-to-market teams to land joint deliverables with the partner.
  • Provide clear and transparent reporting on program status, issues, and risks to executives and stakeholders.

Tools and skills named

Models & research
  • Machine learning2×
  • Inference
Cloud & infra
  • Azure
  • CI/CD
Go to market
  • Go-to-market
  • Partnerships
Operations & finance
  • Program management2×
Product & design
  • Roadmap2×
Ways of working
  • Cross-functional

Words the posting leans on

  • cloud8×
  • engineering8×
  • partner8×
  • technical8×
  • deployment5×
  • execution4×
  • model4×
  • platforms4×
  • program4×
  • coordination3×
  • experience3×
  • joint3×
  • launch3×
  • roadmap3×
  • support3×
  • amazon bedrock2×

Counted from the posting after the mission statement and the legal notices are set aside. The ones near the top are the ones a screener is looking for.

The posting, your resume, and the gaps between them. One click loads all three.

More open at Anthropic

every open role at Anthropic

How this page was made

An automated read of a public job posting, fetched August 25, 2026 and last changed by Anthropic on August 21, 2026. Every list above is pulled from the posting’s own sentences — nothing rewritten, nothing added, no judgment about the role or the company. Counts and seniority are read off the text by rule, so they can be wrong where the posting is unusual. The original is the only thing that binds. Openings close without warning; check the source before spending an evening on it.