Tech Infra Program Manager, DC Power Utilization Management
Posted Updated
About the Role
Amazon Web Services (AWS) is seeking a Technical Infrastructure Program Manager (TIPM) to join the Power & Utilization Management and Planning (PUMP) team within AWS Infrastructure Services. This role is responsible for driving programs that maximize power capacity, improve position planning accuracy, and accelerate rack deployment across AWS's global fleet of data centers.
What You'll Do
As a TIPM on the PUMP team, you will own sub-workstreams within one or more of the following program areas, working under the guidance of senior TIPMs who hold end-to-end accountability. You will drive execution on defined scopes, remove blockers, coordinate across partner teams, and progressively take on broader ownership as you build domain expertise.
Data Center Position Configuration & Automation
The Soapstone workflow produces three interdependent datasets per build — rack placement on the floor plan, power circuit-to-rack mapping (which determines what whips get physically installed), and network cabling plans (which determine how server racks connect to aggregation racks). These outputs determine what gets physically built: general contractors install whips from them, electrical engineers configure three-phase power equipment at the rack level, and mechanical engineers reference them for downstream work. When this workflow is delayed or incorrect, construction stalls, hardware ships to positions that cannot support it, and capacity delivery slips. You will own sub-workstreams within this pipeline — for example, driving configuration for a set of availability zones, managing the intake and validation process through SLATE (the team's build coordination portal), or accelerating Hamlet automation coverage to new building classes (COLO sites, hybrid floor plans, Room 1 configurations).
Power Reclamation & Infrastructure Upgrades
You will drive specific remediation campaigns that recover stranded power across legacy data centers through electrical upgrades, PDC (Power Distribution Cabinet) deployments, cooling improvements, and breaker-level optimizations. The power distribution landscape is complex: four distinct PDC versions are active in the fleet with different breaker counts and ampacity ratings, whip counts range from 18 to 26 per rack depending on product type, and whip lengths are tracked manually in Excel files rather than any system of record. You will coordinate across Field Remediation Engineering (FRE), Construction Project Implementation (CPI), Colocation Regional Engineering (CRE), and equipment procurement teams to execute physical infrastructure changes on time.
Network & Server Position Planning
You will own sub-workstreams that improve how network and server positions are planned, allocated, and made ready for deployment. A key challenge is that every rack product connects to the network differently — connection protocols vary (RoCE vs. EFA), server-to-switch ratios vary (3:1, 2:1, 1:1), and switch rack generations keep evolving (12.8T, 51.2T, 102.4T Ultra). You will work with Fulfillment Planning (FP), Capacity Delivery Planning (CDP), and Global Network Deployment (GND) on initiatives like the Fungible Network program, which deploys a standard product-agnostic network footprint per ROMP so that product flips no longer require cabling rework. You will use and improve tools like RATS and PRP to increase allocation rates and reduce rejection rates at scale.
Power Data Modeling & Metrics
You will contribute to the development and maintenance of power-related data models, dashboards, and sources of truth that inform executive-level decisions on capacity investment, utilization targets, and demand planning. This includes working with data engineering and BI teams to build QuickSight dashboards, validate Power SOT (Source of Truth) data, and support Monthly and Weekly Business Reviews.
Utility Power Management
You will support programs that optimize utility power utilization to meet contractual obligations and avoid financial penalties. This includes monitoring site-level deployed load against contracted capacity, identifying sites at risk of under-utilization, and coordinating with energy and operations teams to improve utilization rates.
Cross-Functional Stakeholder Management
You will work daily with 10+ partner organizations including CDP, CRE, CPI, FRE, DCEO, GND, Hardware Engineering, ML Planning, AISS Science, and Networking. Success in this role requires the ability to influence without authority, drive accountability across organizational boundaries, and escalate effectively when blockers arise.
About the team
The PUMP team sits at the intersection of data center engineering, supply chain operations, and capacity planning. We own the end-to-end lifecycle of how power is planned, configured, allocated, and utilized across thousands of lineups, hundreds of data centers, and dozens of clusters worldwide. Our decisions directly impact AWS's ability to deliver compute, storage, and machine learning capacity to customers — from core EC2 workloads to the latest GPU-accelerated ML platforms like GB300, VR200, and TRN3Max.
Amazon Web Services (AWS) is seeking a Technical Infrastructure Program Manager (TIPM) to join the Power & Utilization Management and Planning (PUMP) team within AWS Infrastructure Services. This role is responsible for driving programs that maximize power capacity, improve position planning accuracy, and accelerate rack deployment across AWS's global fleet of data centers.
What You'll Do
As a TIPM on the PUMP team, you will own sub-workstreams within one or more of the following program areas, working under the guidance of senior TIPMs who hold end-to-end accountability. You will drive execution on defined scopes, remove blockers, coordinate across partner teams, and progressively take on broader ownership as you build domain expertise.
Data Center Position Configuration & Automation
The Soapstone workflow produces three interdependent datasets per build — rack placement on the floor plan, power circuit-to-rack mapping (which determines what whips get physically installed), and network cabling plans (which determine how server racks connect to aggregation racks). These outputs determine what gets physically built: general contractors install whips from them, electrical engineers configure three-phase power equipment at the rack level, and mechanical engineers reference them for downstream work. When this workflow is delayed or incorrect, construction stalls, hardware ships to positions that cannot support it, and capacity delivery slips. You will own sub-workstreams within this pipeline — for example, driving configuration for a set of availability zones, managing the intake and validation process through SLATE (the team's build coordination portal), or accelerating Hamlet automation coverage to new building classes (COLO sites, hybrid floor plans, Room 1 configurations).
Power Reclamation & Infrastructure Upgrades
You will drive specific remediation campaigns that recover stranded power across legacy data centers through electrical upgrades, PDC (Power Distribution Cabinet) deployments, cooling improvements, and breaker-level optimizations. The power distribution landscape is complex: four distinct PDC versions are active in the fleet with different breaker counts and ampacity ratings, whip counts range from 18 to 26 per rack depending on product type, and whip lengths are tracked manually in Excel files rather than any system of record. You will coordinate across Field Remediation Engineering (FRE), Construction Project Implementation (CPI), Colocation Regional Engineering (CRE), and equipment procurement teams to execute physical infrastructure changes on time.
Network & Server Position Planning
You will own sub-workstreams that improve how network and server positions are planned, allocated, and made ready for deployment. A key challenge is that every rack product connects to the network differently — connection protocols vary (RoCE vs. EFA), server-to-switch ratios vary (3:1, 2:1, 1:1), and switch rack generations keep evolving (12.8T, 51.2T, 102.4T Ultra). You will work with Fulfillment Planning (FP), Capacity Delivery Planning (CDP), and Global Network Deployment (GND) on initiatives like the Fungible Network program, which deploys a standard product-agnostic network footprint per ROMP so that product flips no longer require cabling rework. You will use and improve tools like RATS and PRP to increase allocation rates and reduce rejection rates at scale.
Power Data Modeling & Metrics
You will contribute to the development and maintenance of power-related data models, dashboards, and sources of truth that inform executive-level decisions on capacity investment, utilization targets, and demand planning. This includes working with data engineering and BI teams to build QuickSight dashboards, validate Power SOT (Source of Truth) data, and support Monthly and Weekly Business Reviews.
Utility Power Management
You will support programs that optimize utility power utilization to meet contractual obligations and avoid financial penalties. This includes monitoring site-level deployed load against contracted capacity, identifying sites at risk of under-utilization, and coordinating with energy and operations teams to improve utilization rates.
Cross-Functional Stakeholder Management
You will work daily with 10+ partner organizations including CDP, CRE, CPI, FRE, DCEO, GND, Hardware Engineering, ML Planning, AISS Science, and Networking. Success in this role requires the ability to influence without authority, drive accountability across organizational boundaries, and escalate effectively when blockers arise.
About the team
The PUMP team sits at the intersection of data center engineering, supply chain operations, and capacity planning. We own the end-to-end lifecycle of how power is planned, configured, allocated, and utilized across thousands of lineups, hundreds of data centers, and dozens of clusters worldwide. Our decisions directly impact AWS's ability to deliver compute, storage, and machine learning capacity to customers — from core EC2 workloads to the latest GPU-accelerated ML platforms like GB300, VR200, and TRN3Max.
