Overview
As Principal Product Manager for Arm AI tools, you define and deliver the strategy for a unified developer experience—from model to deployment on Arm-based systems. You lead multiple engineering teams, shape partnerships, and drive adoption of a coherent toolset across CLIs, Python, notebooks, IDEs and AI agents. You will accelerate a simple model-to-deployment journey while balancing accuracy, performance and power on edge hardware. This role offers high impact, working closely with ML frameworks and silicon partners to scale Arm’s AI tooling.
Pay / Benefits
- sponsorship for skilled workers
- accommodations during recruitment
- hybrid working arrangements
- equal opportunities employer
Responsibilities
- Set product vision, roadmap and success metrics for Arm AI tools
- Define developer workflows and ensure adoption across ML engineers and application developers
- Coordinate across engineering teams to align architecture and product investments
- Develop ecosystem partnerships with ML frameworks, runtimes, silicon partners and accelerators
- Deliver an exceptional developer experience across CLIs, Python, notebooks, IDEs, CI and AI agents
- Prioritize capabilities improving model compatibility with Arm accelerators and edge deployment
- Translate UX research and market data into roadmap decisions
- Explore agentic model optimization approaches and reproducible automation
Key requirements
- Proven product management experience in development tools (CLIs, SDKs, APIs)
- Knowledge of ML frameworks and runtimes (PyTorch, ONNX, ExecuTorch, Llama.cpp, vLLM, LiteRT)
- Understanding of model deployment to edge/embedded/heterogeneous computing and trade-offs in accuracy, performance, power
- Clear communication with developers, partners and senior leaders
- Ability to align multiple stakeholders without direct authority
- Good judgment on build, partnerships and acquisitions
- Model formats, quantization, graph partitioning
- Latency and memory optimization for hardware acceleration
- Experience with developer tooling and agent-based workflows
…
