About the Company

We're a small team solving one of the hardest problems in human-computer interaction: a noninvasive wearable that turns thought into text, no surgery required.

By pairing ultra-high-density neural sensing with our Brain Foundation Model, we decode neural signals with a fidelity once reserved for implants. Our mission is to give a billion people a direct link between mind and machine - expanding how humans think, communicate, and create.

We are building a next-generation AI companion wearable powered by EEG/BCI technology. Our device reads and responds to neural signals in real time, creating a deeply personal experience that adapts to each user. This is not another productivity tool or a gadget — it is a new category of technology built around human potential.

We are backed by strong investors, moving fast, and assembling a world-class team to bring this to market. If you thrive at the frontier of what’s possible and want to build something that genuinely changes how people interact with their own minds, we want to talk to you.

About the Role

You will architect and own the firmware for our wearable’s distributed-compute platform: a media and connectivity MCU running camera capture, audio runtime, Wi-Fi streaming, and onboard storage; a sensor-acquisition MCU running the biopotential signal chain and BLE streaming; and an always-on MCU enforcing power management, privacy, and thermal control.

The firmware is the connective tissue of our wearable — every subsystem either lives inside it or talks to it. You will report to the Head of Hardware, with regular exposure to Sabi’s CEO and CTO, and partner closely with the EE, audio, camera, AI/ML, and reliability leads.

Each subsystem lead provides their algorithms and interface specs; you provide the runtime, drivers, and orchestration that make the whole platform work. You will also work extensively with external engineering teams at our contract manufacturers — the work is outsourced; the accountability is yours.

What You’ll Do

Architect firmware across the distributed-compute platform: RTOS choice, task structure, inter-processor protocols, OTA, and time sync.

Own the runtime that hosts every subsystem, audio DSP, camera capture pipeline, biopotential acquisition, through specified interfaces with each subsystem lead.

Own the media and connectivity MCU pipeline, camera capture, audio runtime, wake-word, Wi-Fi streaming, and onboard logging, concurrently within strict CPU and memory budgets.

Drive low-power firmware on the always-on MCU: state machines for standby, assist, and continuous modes; hardware-enforced privacy; thermal throttling.

Build the multi-MCU time-sync layer that lets us correlate EEG, audio, and camera data downstream.

Establish the firmware engineering practices that scale: build and release pipelines, on-device telemetry, automated test, OTA with safe rollback, field debug tooling.

Partner with the EE lead on hardware bring-up and boot path; with the reliability lead on field telemetry, error handling, and diagnostic surfaces.

Bring up ASICs in collaboration with the EE and silicon teams

Ship the product by the end of year, and build and lead the firmware team as we scale to production.

What We’re Looking For

Must-Haves

10+ years of embedded firmware engineering, with at least one shipped consumer product where you owned firmware architecture end-to-end.

Deep expertise across embedded RTOSes and bare-metal ARM Cortex-M, with familiarity across at least two ecosystems (e.g., Zephyr, FreeRTOS, ESP-IDF, ThreadX, NuttX).

Hands-on experience hosting real-time DSP runtimes alongside wireless connectivity on resource-constrained MCUs — integrating algorithms owned by other teams.

Strong background in multi-radio coexistence (Wi-Fi + BLE), low-power state-machine design, and OTA with safe rollback.

Comfortable in the lab with JTAG/SWD, logic analyzers, and protocol sniffers — able to drive bring-up from first power-on through end-to-end functional demos.

Nice-to-Haves

Deploying neural network inference to low-power MCUs or dedicated AI accelerators — model conversion, quantization, runtime integration.