Staff Technical Program Manager, Physical AI

Apptronik•Sunnyvale, CA
•$240,000 - $280,000

About The Position

Apptronik is seeking a highly technical Staff or Principal Technical Program Manager to lead programs that train Apollo's AI. The role owns the full physical AI training lifecycle, including training compute infrastructure, the data engine, pretraining, mid-training and post-training runs, and deployment of models onto robots in the field. The TPM will work as a technical peer to research and infrastructure leadership, set strategy for planning and resourcing training programs, oversee cloud and on-premises GPU capacity, and translate research goals into executable plans with clear gates and decisions. As the function grows, this person will define program management practices that help Apptronik train and ship physical AI models faster.

Requirements

  • Bachelor's degree in Computer Science, Electrical or Computer Engineering, or a related technical field.
  • 10+ years of experience in technical program management, ML engineering, or ML infrastructure, including leadership of large-scale training or compute programs at a senior, staff, or principal level.
  • Deep experience leading large-model training programs, including pretraining, mid-training, fine-tuning, and distillation, and fluency in planning, resourcing, monitoring, and recovering training runs.
  • Hands-on understanding of GPU clusters in cloud and on-premises environments, including schedulers, distributed training, high-speed interconnects, and storage; experience driving capacity planning, procurement, and cluster bring-up.
  • Ability to read training code, interpret loss curves and evaluation results, and diagnose data and infrastructure issues alongside technical teams.
  • Comfort with Python and a modern framework such as PyTorch or JAX.
  • Experience with physical AI training data, including teleoperation, demonstration capture, simulation and synthetic data, annotation, and quality control at scale.
  • Experience building ML model evaluation frameworks and release criteria, ideally spanning simulation and real-world testing.
  • Familiarity with taking models from training to on-device inference, including optimization for embedded and edge AI hardware.
  • Proven success managing cloud providers, hardware vendors, and data partners, including negotiating capacity and resolving cross-organizational dependencies.
  • Ability to plan research programs around uncertainty using experiments, decision points, and gates while holding infrastructure and data work to firm commitments.
  • Proven ability to allocate scarce compute and data across competing teams, align research, engineering, and executive leadership, and make prioritization decisions.
  • Ability to translate company strategy into compute, data, and model roadmaps and influence those roadmaps using data-driven insight.

Nice To Haves

  • Experience at a frontier AI lab, hyperscaler, or robotics foundation model company.
  • Experience standing up or scaling GPU clusters or data center capacity.
  • Direct experience with robot learning, vision-language-action models, or large-scale robot data programs.
  • Master's or PhD in Computer Science, Machine Learning, or a related field.

Responsibilities

  • Own the strategy and execution of training compute across cloud and on-premises GPUs, including capacity forecasting, build-versus-buy decisions, vendor and cloud partnerships, cluster bring-up, and compute utilization, cost, and allocation.
  • Manage pretraining, mid-training, fine-tuning, and distillation programs, defining compute budgets, data freeze dates, checkpoint and evaluation gates, and go/no-go criteria; lead responses to run divergences and infrastructure failures.
  • Lead the data program that translates model needs into training data from teleoperation, human demonstrations, and simulation, including collection plans, quality standards, SOPs, data throughput, pipelines, and storage.
  • Establish evaluation and model release processes across simulation and real robots, including statistically sound benchmarks, regression gates, and evidence-based promotion criteria.
  • Own the path from trained checkpoint to on-robot inference, including model optimization for edge hardware and integration with software, controls, and firmware.
  • Build a fleet-learning loop that returns fleet data to post-training and enables safe rollout of improved policies.
  • Own program execution end to end, set strategic direction, identify resource requirements, and deliver on company goals and roadmap.
  • Challenge organizational assumptions and foster a results-driven culture.
  • Communicate program status, risks, trade-offs, and decisions to executive stakeholders, and clarify requirements, goals, and milestones across teams.
  • Mentor TPMs and engineers, define frameworks, tooling, and operating rhythms, and raise the technical bar of program management.
© 2026 Teal Labs, Inc
Privacy PolicyTerms of Service