СШАData Center Network Architect
1 мес. назад
175k–230k USD / годСШАSeniorOnsite
network architecturenetwork deploymentnetwork automationtroubleshootingperformance optimizationsecurity
Lead design, deployment, and operation of networks connecting distributed compute infrastructure for AI workloads.
О компании
- Headquartered in San Francisco’s vibrant SoMa neighborhood, we are an eclectic group of creative thinkers who value open communication, teamwork, and a ‘make it happen’ approach to addressing complex challenges.
- embraces diversity and equal opportunity in a serious way. We are committed to building a team that represents a variety of backgrounds, perspectives, and skills.
- We’re hiring talented individuals who are driven by success and are passionate about shaping the future of renewable energy. If that sounds like you, we’d love for you to consider joining the rapidly growing team at .
Обязанности
- We are looking for a Network Architect to own the design, deployment, and operation of the network connecting XFRA’s distributed compute fleet.
- Your scope will high-performance networking within GPU clusters, north-south connectivity between compute sites and customers, and the wide-area architecture connecting our fleet to cloud platforms and compute partners. You will turn workload requirements into practical designs that balance performance, availability, security, deployment speed, and cost.
- This is a hands-on technical leadership role. You will select and configure equipment, qualify ISP services, automate deployments, troubleshoot production issues, and establish the standards that allow our network to scale. You will work closely with compute, orchestration, security, deployment and commercial teams.
- Own the network architecture. Develop reference designs for residential nodes and larger commercial GPU clusters, including routing, switching, security gateways, addressing, management access, physical connectivity, and capacity planning.
- Engineer data center connectivity. Design north-south paths for customer traffic, model distribution, storage access, and fleet management, alongside east-west fabrics for communication between servers. Evaluate high-speed Ethernet, InfiniBand, and RoCE against workload requirements, including bandwidth, latency, congestion, and failure behavior.
- Select and qualify fiber services. Evaluate shared PON broadband, dedicated internet access, Ethernet private lines, wavelength services, and dark fiber where appropriate. Assess committed versus advertised bandwidth, oversubscription, symmetry, latency, jitter, packet loss, public addressing, routing options, service guarantees, repair commitments, lead times, and total cost.
- Build provider relationships. Work directly with ISPs and carriers on service availability, technical requirements, provisioning, acceptance testing, and incident escalation. Verify physical path diversity and shared dependencies when designing redundant connectivity.
- Optimize the distributed fleet. Define when sites should connect directly to the internet, through regional aggregation, or through private interconnects. Evaluate transit, peering, cloud connectivity, traffic engineering, and failover options using measured performance and economics.
- Partner with orchestration engineering. Provide network topology, capacity, and health signals that support workload placement, traffic steering, and recovery. Account for model transfers, request traffic, and distributed inference communication when determining which workloads can run efficiently within or across sites.
- Build security and resilience into deployments. Implement tenant isolation, separation of compute and management traffic, encrypted connectivity, access controls, and DDoS mitigation with the security team. Design remote recovery and appropriate out-of-band access for unattended sites.
- Automate and operate at scale. Build repeatable provisioning, configuration validation, staged changes, rollback, and inventory management. Establish telemetry, service objectives, alerting, runbooks, and acceptance tests; lead network incident diagnosis and corrective actions.
Условия
- ⚡ Competitive compensation + equity grants at a well-funded, venture-backed company
- ⚡ Comprehensive benefits: 100% employee premiums for base plans on medical, dental, vision with options for additional coverage. Parental leave up to twenty four (24) weeks depending on eligibility
- ⚡ Comfortable, sunny office space located near BART and Caltrain public transit
- ⚡ Strong focus on team building and company culture: Employee Resource Groups, monthly social events, SPANcakes recognition breakfast, lunch, and learns
- ⚡ Flexible hours, one holiday per month, and flexible time off
- Interested in joining our team? Apply today and we’ll be in touch with the next steps!
Другое
- Decarbonization is the process to reduce or remove greenhouse gas emissions, especially carbon dioxide, from entering our atmosphere.
- Electrification is the process of replacing fossil fuel appliances that run on gas or oil with all-electric upgrades for a cleaner way to power our lives.
- Enabling homes and vehicles powered by clean energy
- Making electrification upgrades possible
- Building more resilient homes with reliable backup
- Designing a flexible and distributed electrical grid
- 8+ years of relevant network engineering experience, or equivalent demonstrated expertise, including ownership of production architecture and hands-on implementation.
- Strong experience with both data center networks and WAN or ISP connectivity across multiple sites.
- Deep understanding of TCP/IP, BGP, IPv4/IPv6, routing, switching, segmentation, VPNs, firewalls, QoS, and high-availability design.
- Practical understanding of fiber access and carrier service models, including how to validate real performance, availability, and commercial service commitments.
- Experience designing and troubleshooting high-speed server networks, including NICs, optics, cabling, MTU, congestion, and oversubscription.
- Ability to translate application traffic patterns and service objectives into network requirements, and explain performance, resilience, and cost tradeoffs clearly.
- Hands-on Linux troubleshooting and network automation using Python, Ansible, or comparable tools, with version-controlled configurations and disciplined change management.
- A record of taking ambiguous problems from design through deployment and production support, working effectively with engineering teams, vendors, and field operators.
- Production experience with GPU clusters, AI inference, HPC, InfiniBand, RoCE, RDMA, or NCCL performance analysis.
- Experience with EVPN/VXLAN, leaf-spine fabrics, cloud interconnects, internet peering, or operating an autonomous system.
- Experience operating a large fleet of unattended edge sites, including zero-touch provisioning, carrier-grade NAT constraints, and remote recovery.
- Familiarity with Kubernetes networking, multi-tenant compute platforms, and network-aware workload scheduling.
- Experience qualifying carrier services and negotiating technical requirements across multiple providers or regions.