Network Engineer Jobs in San Francisco, CA

The Role

The network is where AI verification becomes real. Lucid Computing, in partnership with the Research Institutes of Sweden (RISE), is building and operating an adversarial testbed for developing and red-teaming AI compute verification solutions. The premise of the project is that verification claims — where AI compute runs, how it is used, what model is loaded — are only as good as the adversarial attacks they can withstand. Our job is to build verification solutions that can survive such attacks. Network devices and network code form the backbone of these solutions, including designs like:

• Traffic shaping for workload classification — bandwidth-constrained pod uplinks, telemetry aggregation, and randomized inference routers/load balancers designed to make covert frontier training economically infeasible, without inspecting or retaining traffic contents.

• Network taps and the evidence plane — line-rate traffic capture and hashing, cryptographic commitment of network telemetry (rather than traffic contents), and the software that lets programmable NICs (FlexNICs) and traffic shapers constrain and observe the I/O path of GPU servers.

• Assurance and receipt infrastructure — software that cross-checks network telemetry against signed job manifests, evaluates policy, and emits chained, verifiable receipts.

The devices you build must remain trustworthy when the adversary is the operator of the network itself — potentially a well-resourced, nation-state-class actor. Requirements like "never mirror or retain full traffic," "commit only to pre-agreed evidence," and "assume every component around you may be hostile" reshape familiar network designs in unfamiliar ways, and many of the right answers haven't been invented yet. You'll help invent them: contributing to threat models, proposing new security protocols and device designs, defending them in design reviews, and then hardening and remediating them when the red team publishes what it broke.

You'll work alongside hardware/FPGA engineers (DC-SCM, FlexNIC prototypes), security engineers, and verification researchers, and your work will directly shape open, published reference architectures intended to be reused by frontier labs, governments, and future compute agreements.

What We're Looking For (Required)

• Real-world experience writing production software for network devices (e.g., switches, routers, NICs/SmartNICs, load balancers, firewalls, taps/packet brokers) and/or network-focused software for large-scale network-based applications (e.g., VPNs, traffic management, load balancing, packet processing, or telemetry systems serving significant real-world traffic).

• Hands-on network security experience — this is a hard requirement, and depth here is heavily weighted. Examples: secure device design, hardened network protocols, traffic analysis and anomaly detection, DDoS mitigation, network access control, or security-focused code and design review of networked systems.

• Strong networking fundamentals: L2–L4 protocols, routing and switching behavior, network topologies, load balancing, QoS/traffic shaping, and performance characteristics of real networks under load.

• Strong systems programming skills in languages appropriate to the data plane and control plane (e.g., C, C++, Rust, Go) and deep familiarity with the Linux networking stack.

• A track record of building robust, reliable software at scale — the bigger and more demanding the networks and devices you've shipped for, the better.

• Creative, outside-the-box thinking. You should be excited (not exhausted) by problems with no established playbook, comfortable reasoning about adversaries, and able to originate and defend novel protocol and device designs — then revise them honestly when the red team proves you wrong.

• Ability to work effectively in a small, fast-moving team spanning software, hardware, and security, and to communicate designs clearly in writing.

Nice-to-Haves (the more, the better)

• Experience developing network device hardware (switch/NIC/appliance hardware, embedded network systems, or close collaboration with hardware teams on device bring-up) — a strong plus.

• Experience physically securing network devices — tamper-evident/tamper-resistant design, secure installation and cabling assurance, physical security assessments of network infrastructure — a strong plus.

• Data-plane programming: DPDK, XDP/eBPF, P4, or programmable switch/SmartNIC/DPU platforms.

• FPGA experience: working with or writing software for FPGA-based network appliances, or comfort collaborating with RTL/firmware engineers.

• Applied cryptography and attestation: hashing/signing at line rate, hash-chained logs, certificate/key management, TPM/TEE attestation flows.

• Offensive security or red-team experience with networked systems; familiarity with covert channels and side channels in networks.

• Familiarity with AI/ML infrastructure: GPU cluster networking (InfiniBand/RoCE, NVLink-scale fabrics), inference-serving architectures, or the traffic patterns of training vs. inference workloads.

• Experience with telemetry and observability systems: packet capture at line rate, flow monitoring, high-throughput time-series pipelines.

• Contributions to open-source networking projects or open standards (IETF, OCP, etc.).

• Interest in AI safety, security, and governance — our work exists to make high-stakes agreements about AI verifiable, and caring about that mission helps.

What You'll Get

• Ownership of a critical layer of a first-of-its-kind system: verification infrastructure installed in a production-grade European datacenter and adversarially tested in public.

• Your designs and code published as part of open reference architectures used across the AI verification field.

• An elite, mission-driven team, and direct collaboration with an independent government-backed research institute.

• Competitive compensation and benefits program.

Compensation Range: $100K - $450K