Private Draft

The 29 personas behind AI

We’ve organized every stage and persona in the AI supply chain, informed by real recruiting at frontier companies. Click any row to see matching profiles from our talent graph.

Shaped by Industry Experts

← ATOMS & ENERGYUSERS & MARKETS →

← Back

Program Management

Sequences capability creation

Program Management

Summary

Known as: Technical Program Manager, Program Manager (AI/ML), Research Program Manager, TPM (Research & Training), TPM (Infrastructure)

The connective tissue that keeps AI orgs executing at scale. At frontier labs, TPMs are strategic partners to research leads — co-developing execution plans, structuring tradeoffs across research bets, post-training recipes, infrastructure capacity, and regression risk. At applied-AI companies, the work centers on ML model productionization and cross-team delivery for ML-powered features. The scope has expanded well beyond traditional project tracking: the best program managers in AI own sequencing and operational governance of model development.

Specializations

Model Development & Launch Programs — Partners with research leads to sequence capability creation, manages training run schedules and eval-gated milestones, and drives launch readiness including eval sign-off workflows, regression thresholds, and go/no-go decisions. Named to signal the full lifecycle from research planning through launch gating. The operational backbone of model development at frontier labs.

Infrastructure Programs — Coordinates hardware, networking, and compute capacity across training and serving. Manages cluster bring-ups, capacity allocation, vendor timelines, and the physical logistics that underpin model development. At orgs building custom hardware, extends into full hardware program management: getting new designs from prototype through factory readiness and into volume production.

Applied ML Programs — Drives ML experiment coordination, model productionization, and cross-team delivery for ML-powered features. The bulk of TPM hiring volume — at companies using ML in products (ads, recommendations, search, content ranking) rather than building foundation models.

Where the Work Lives

[1]Substrate

[2]Compute

Secondary

Coordinates hardware capacity, cluster bring-ups, and vendor timelines for infrastructure programs.

[3]Intelligence

Primary

Sequences research bets, manages training run schedules, and drives eval-gated milestones.

[4]Systems

Secondary

Drives launch readiness, regression thresholds, and go/no-go decisions across model development.

[5]Distribution