Senior Embedded AI Engineer
Seoul, South Korea • Full-time
Posted 1mo ago
About the job
Sonatus is seeking an Embedded AI Engineer to join our Korea Customer Success team, focusing on deploying and managing AI capabilities within vehicles. This role involves integrating our Fastlane™ Edge AI runtime onto customer hardware, onboarding AI models from various sources, and assisting customer engineers in developing their own models using the Sonatus SDK. You will be responsible for validating models on actual vehicles, tuning virtual-sensor models with customer data, and serving as the primary technical contact for customers on edge AI integration. This is a hands-on position requiring comfort with Python, Linux containers, automotive ECUs, and on-vehicle testing, with model development always in service of customer deliverables.
Responsibilities
- Integrate and maintain the Fastlane Edge runtime on customer hardware, including CAN/CAN-FD, DBC/ARXML, sandbox deployment, security policies, and cloud connectivity.
- Integrate the runtime with customer platforms using the Sonatus Southbound SDK when direct bus access is unavailable.
- Maintain Korea integration runbooks and default configurations, contributing fixes to the platform team.
- Validate, package, and deploy AI model bundles from partners, customers, and Sonatus using the Sonatus ADK, including input-signal mapping, resource budgeting, and OTA deployment.
- Verify multi-model concurrency on a single ECU within specified resource envelopes and document per-model resource profiles.
- Migrate Python-developed models to production-ready native C++/ONNX Runtime bundles, optimizing them and validating numerical parity.
- Deliver SDK/ADK training and support to customer engineers authoring their own models.
- Capture and document customer AI workload requirements, including use cases, accuracy/latency targets, and resource constraints.
- Treat documentation as code, creating integration guides, FAQs, and release notes in Korean and English.
- Process customer vehicle captures through the Sonatus MLOps toolchain for data ingest, training, model comparison, ONNX export, quantization, and latency measurement.
- Tune Sonatus virtual-sensor and estimation models on customer data, delivering validated bundles with reports.
- Define ground-truth capture and dataset versioning, contributing to toolchain improvements.
- Execute on-vehicle validation drives, contribute to acceptance criteria, and produce traceable test evidence.
- Provide on-site support for test events and demos, troubleshooting integration issues.
- Act as a technical contact for edge AI integration issues, triaging and tracking customer-found defects.
- Leverage AI tools like Claude and GitHub Copilot to accelerate workflow tasks within data governance constraints.
Requirements
- Experience with Python training pipelines, Linux containers, and automotive ECUs.
- Familiarity with on-vehicle testing environments.
- Experience with model development in the context of customer deliverables.
- Experience with hardware bring-up and integration on automotive targets (e.g., CAN/CAN-FD, DBC/ARXML).
- Experience with SDK integration.
- Experience with AI model onboarding, validation, and packaging.
- Experience with concurrency verification and resource profiling on embedded systems.
- Experience migrating models from Python to native C++/ONNX Runtime.
- Experience with automated numerical-parity validation.
- Experience delivering technical training and support.
- Experience capturing and documenting technical requirements.
- Experience with MLOps toolchains, including data ingest, training, model conversion (ONNX), quantization, and latency measurement.
- Experience tuning virtual-sensor or estimation models.
- Experience defining ground-truth capture and dataset versioning.
- Experience with on-vehicle validation and acceptance criteria.
- Experience troubleshooting integration issues on-site.
- Experience triaging and tracking defects.
- Familiarity with AI tools such as Claude and GitHub Copilot.
