Systems Engineer, Transaction Landing

Temporal
· New York City (preferred) or US Remote
· Full-time

Temporal builds low-latency infrastructure at the core of Solana. We operate across transaction landing, validator infrastructure, and proprietary trading systems. Our flagship product is Nozomi, for transaction landing.

This is not a middleware or dashboard role. We work close to the metal: networking, scheduling, memory layout, validator clients, and production trading infrastructure.

We are hiring a systems engineer to own our transaction landing service end to end. The backend, the networking, and the infrastructure it runs on.

A customer hands us a transaction. We get it into a block as fast and as reliably as physically possible. Everything between those two points would be yours. You would be the primary owner of that surface, not a contributor to it: the person accountable for its latency, its correctness, and its uptime.

The request path is written by hand rather than assembled from a general-purpose async stack, because the frameworks cost more microseconds than we are willing to spend. Underneath it sits a dispatch layer that decides how to get each transaction in front of the right validator in time, on a budget measured in microseconds. All of it runs on bare metal across many regions.

This role is for someone who wants a large, real system to own outright, and who gets uncomfortable when they cannot explain a p99 down to the syscall. Years of experience matter less than slope: what we care about is taste, speed, ownership, and the ability to become dangerous in a complex codebase quickly.

• Own the service end to end: backend, networking, and infrastructure, through design, implementation, deployment, operation, and on-call.

• Cut latency and raise landing rates on paths where the difference between good and great is measured in microseconds.

• Profile and optimize across the full stack: syscall behavior, memory layout, cache effects, lock-free data structures, allocator behavior, queue design, CPU scheduling, NUMA, packet budgets, and tail latency.

• Improve the network path between our infrastructure and validators, including placement, peering, routing, and transport choice.

• Operate a global bare-metal fleet: provisioning, deploys, pinning and isolation, kernel and NIC tuning, capacity, and the tooling that makes all of it repeatable.

• Build the observability to know, per transaction, where the time went and why something did not land. Those numbers drive customer reporting and billing, so they have to be right.

• Own correctness where it touches money: deduplication, revert protection, fee and tip accounting, and reconciliation against chain state.

• Investigate production incidents where the answer may involve kernel behavior, packet drops, a NIC queue, lock contention, a validator's own behavior, or a Solana runtime edge case.

• Exceptional Rust. You are comfortable in unsafe code when it is warranted, you know what the compiler is doing, and you can reason about ownership, allocation, and atomics without hand-waving. Equivalent depth in C or C++ with real Rust intent is fine.

• Fluency with low-level primitives: syscalls, epoll and , sockets and socket options, batched send and receive, shared memory, lock-free queues, atomics and memory ordering, cache lines and false sharing, CPU pinning, and NUMA.

• Real experience building or optimizing low-latency, high-throughput production systems where tail latency was the metric that mattered.

• Deep Linux performance skill: perf, eBPF, flame graphs, hardware counters, packet captures, and tracing, plus the judgment to know which one answers the question in front of you.

• Networking depth: UDP and QUIC behavior, TCP and TLS internals, kernel bypass, NIC and driver behavior, routing, and how packets actually move between datacenters.

• Comfort operating infrastructure, not just writing services. Bare metal, systemd, configuration management, deploys, and being the person paged when it breaks.

• Strong ownership: you can take an under specified production problem, investigate it, build the fix, ship it, and operate it.

Especially relevant: HFT or market-making infrastructure such as feed handlers, order gateways, and colocation; kernel bypass and fast-path networking including , DPDK, , RDMA, or Solarflare and Mellanox stacks; HPC, distributed databases, storage engines, schedulers, compilers, runtimes, or operating systems; validator clients, Solana internals, transaction propagation, or MEV; SRE work for latency-sensitive bare-metal fleets.

You do not need years of blockchain experience, a crypto background, a specific degree, or years of seniority. Exceptional early-career engineers with the right fundamentals and hunger are encouraged to apply. We are happy to talk to engineers from HFT, infra, databases, kernels, compilers, trading systems, HPC, and networking.

Temporal is small, technical, and fast-moving. Engineers own real systems, not tickets. High…