Member of Technical Staff (TPM, Compute)

Department: Platform & Infrastructure

Location: San Francisco

Compensation: $170K – $265K • Offers Equity

Employment Type: FullTime

Perplexity is looking for a technical program manager to own how we plan, bring online and run the GPU compute behind every product.

Perplexity serves Ask, Computer and API traffic on GPU capacity across multiple cloud and neocloud providers, and demand shifts with every new customer, product launch and model release. Working across engineering, research, Finance and providers, you’ll forecast how much compute we need and when, bring new capacity into production, and own how the fleet performs today, driving fixes when reliability, utilization or cost fall short. The ideal candidate knows GPU fleets, handles technical and commercial details, and is energized by building a function with little precedent.

Our Mission

Perplexity’s mission is to power curiosity. Curious people are the people who drive change in the world. Driving change is a continuous cycle of learning, building, and integrating.

Learn: curious people constantly learn new things by asking more. They question the status quo in their own expertise and they constantly learn outside of it. Research is essential to them and never ending.

Build: curious people make and create things, to show the world their new answers to problems no one else ever questioned. They take action on what they’ve learned. Makers need tools to create their products, their companies, their reality.

Integrate: they must interact with the world as it is to drive change and adoption. True leaders do not simply build something and hope. They must have armies of agents and workers who can constantly work in millions of small ways.

What You’ll Do

• Partner with Finance on the compute demand forecast, 3–12 months out, driven by customer growth, model releases, and post-training and evaluation plans rather than trend lines

• Turn the forecast into decisions with Finance and engineering leadership on commitments and budget, checking that new capacity fits the plan before we commit

• Bring new capacity online and lead transitions, from contract to production, including provider onboarding, acceptance criteria, migration and decommissioning

• Build a single, trusted view of the fleet: capacity, cost, allocation, and uptime, reliability and utilization by provider and workload

• Drive daily utilization, reliability and cost-efficiency as core metrics, reclaiming idle capacity and surfacing trade-offs rather than letting them become side effects

• Own inference and compute requests end to end: triage and resolve GPU, deployment and quota requests from teams across Perplexity, and build self-serve paths so routine requests no longer need a person in the loop

• Manage compute provider relationships, with scorecards and regular reviews of reliability, support and delivery, and use that data to shape how we work with providers over time

• Partner with engineering, research, and finance teams on capacity decisions that match the different needs of training, batch inference and real-time production inference

• Build the operating system for compute: the processes and tooling to request, plan, allocate and track compute, with clear owners

Qualifications

• 6+ years of technical program management, capacity planning or infrastructure operations in cloud environments, leading programs across teams and vendors

• Direct experience with GPU or accelerator fleets across training and inference workloads

• Experience working directly with cloud or neocloud providers on reserved capacity, capacity commitments, onboarding or vendor performance management

• Experience with infrastructure cost modeling, unit economics or cost attribution in partnership with Finance

• Technical depth in cluster management, scheduling and orchestration, and the data skills to build forecasts and dashboards yourself

• The judgment to surface hard trade-offs and communicate them clearly from engineers to executives

• Thrives in a small, agile team; has initiative and desire for ownership without much precedent to lean on