Lattice Edge Runtime and Sensor Perception at Anduril
Anduril hardware units operate in austere environments where network bandwidth is intermittent and hardware failures occur regularly. Edge nodes process multimodal sensor feeds from cameras, radar units, and RF receivers directly on bare-metal systems. Rust is our primary language for low-level sensor ingest and hardware interaction. Memory safety prevents segfaults during field operations. Stability is vital. The edge runtime uses zero-copy ring buffers and Tokio task pools to ingest sensor frames, run SIMD pre-processing, and dispatch tensors to local GPU accelerators. Fast execution is critical. Microseconds count. You will write deterministic systems code in Rust that processes high-frequency sensor streams with predictable tail latency, ensuring critical track updates propagate to Lattice OS within milliseconds.
Base compensation ranges between $191,000 and $253,000 USD, supplemented by equity grants. Clear bands.
Core Engineering Responsibilities
- Write low-latency Rust sensor ingestion services interfacing with camera sensors, radar, and RF hardware.
- Build zero-copy memory ring buffers feeding vectorized image batches to edge GPU inference pipelines.
- Optimize asynchronous task scheduling in Tokio to maintain sub-10ms sensor processing deadlines. Deterministic concurrency needed.
- Profile memory bandwidth, CPU cache utilization, and thermal throttle limits on embedded Linux platforms.
Technical Requirements
- Proven track record building real-time systems or embedded platforms in Rust or C++.
- Experience optimizing memory pipelines with SIMD intrinsics or GPU shared memory architectures.
- Comfortable diagnosing hardware-level bottlenecks, IPC overhead, and Linux scheduler jitter.
- Disclosed compensation: $191,000 to $253,000 USD base salary plus equity.
