Senior AI / Embedded ML Engineer

Your mission

Build edge ML systems that are accurate, fast, and efficient on small devices. Improve model accuracy, inference latency, memory footprint, and power use with every release. Turn sensor data into dependable product features. Move quickly from prototype to field-tested deployment and make decisions with limited data. Build a culture that ships working ML on hardware and grow a team that delivers on time.

What you will own

Data ingestion and pipelines. Design and build data ingestion pipelines from IMUs, accelerometers, gyroscopes, microphones, and other environmental sensors. Clean, label, synchronize, and store raw sensor data. Build tools to collect, version, and manage training datasets at scale.
Model development and training. Develop and train models for classification, regression, anomaly detection, and signal processing. Select the right architecture for each problem and hardware target. Fine-tune pre-trained models for domain-specific data, and run experiments to compare performance.
TinyML and embedded deployment. Optimize models for ARM Cortex-M, RISC-V, and DSP targets using quantization, pruning, and knowledge distillation. Deploy with TensorFlow Lite Micro, Edge Impulse, ONNX Runtime, or ExecuTorch. Integrate inference into C, C++, or Rust firmware, and profile memory, power, and real-time performance.
Hybrid LLM integration. Design hybrid architectures that combine on-device models with LLM-based reasoning. Build pipelines that route tasks between edge inference and cloud or edge-hosted LLMs. Evaluate tradeoffs in latency, accuracy, and power.
Embedded software and systems integration. Write clean, well-tested embedded software that puts ML inference into real-time systems. Work with FreeRTOS, Zephyr, and bare-metal firmware. Co-optimize the full system stack with hardware and firmware teams.
Documentation and reporting. Document design decisions, pipeline configurations, model benchmarks, and deployment procedures. Prepare technical reports and presentations for internal teams and stakeholders. Bring relevant TinyML and edge AI advances into the team.
Bring-up and field support. Provide technical support during hardware bring-up, system integration, and field testing. Take part in design reviews and give useful feedback across the stack.
Lasting corrective action. Contain field and model issues fast. Find root causes, such as data drift or sensor noise, and verify that fixes prevent recurrence.
Measurable results. Own clear, tracked targets for each model and device. Meet accuracy targets on real sensor data. Keep inference latency, memory footprint, and power per inference within budget. Ship models that pass field testing on the first attempt where possible. Keep the training data versioned and reproducible.

What you bring

Strong background in signal processing, sensor data handling, and real-time system constraints.

Hands-on experience with IMUs and other sensors, including accelerometers, gyroscopes, barometers, and microphones.

Proficiency in Python for ML development with PyTorch, TensorFlow, or scikit-learn.

Experience with C or C++ for embedded systems development.

Solid understanding of model optimization, including quantization, pruning, and distillation.

Experience deploying models with at least one embedded ML framework, such as TFLite Micro, Edge Impulse, or ONNX Runtime.

Strong understanding of memory-constrained and power-constrained environments.

Excellent problem-solving skills, with the ability to work independently and as part of a team.

A degree in electrical engineering, computer engineering, computer science, or a related field, or equivalent practical experience.

A plus but not required

Experience with FreeRTOS or Zephyr, familiarity with MCU families such as NXP, STM32, or ESP32, experience designing hybrid edge-LLM pipelines or running small language models on device, background in feature extraction (FFT, filter banks, wavelet transforms), experience with hardware-aware neural architecture search or AutoML for edge targets, familiarity with Rust for embedded systems, and prior work on wearables, robotics, industrial sensing, or IoT products.

What success looks like

Models run accurately on real devices within their memory and power budgets. Sensor data flows reliably from device to dataset. Hybrid ML systems route work well between the edge and the cloud. Field issues surface early and get fixed quickly. Your teammates grow, and the team ships edge ML that works in the field.