Machine Learning Engineer, Radar Systems

This machine learning engineering role is part of Idein’s AI Development Division and focuses on commissioned development for a client with expertise in radar systems and product development. The project aims to create radar-based solutions by combining the client’s domain knowledge with Idein’s machine learning expertise. The work covers the process from technical research and experimentation to development intended for operation on real devices.

Responsibilities include researching and reproducing signal-processing and machine-learning algorithms for radar data, investigating methods for domain-specific problems, implementing approaches, and conducting evaluation experiments. The engineer will inspect and organize received data, plan data collection, study and implement methods based on research papers, and develop original approaches. Other activities include interviewing the client about business processes, documenting the work in writing, summarizing findings in slides, designing models, managing experiments, consolidating evaluation results, and organizing next actions. Regular meetings and ongoing discussions with client engineers are used to establish technical direction and determine how machine learning should be applied.

The position is intended for development that considers domain operations and physical data acquisition, rather than only routine model training and validation. Algorithm design may require consideration of sensor characteristics, the mathematical aspects of signal processing, and practical constraints. The project is in its launch phase, so the engineer is expected to contribute to technical policies and development processes, make decisions and proposals based on hypotheses in unfamiliar areas, respond flexibly to sudden changes in direction or specifications, and lead tasks with a high degree of uncertainty.

During the first three months, the expected progress includes learning project terminology and domain knowledge, understanding the overall project and current client issues, and building working relationships with team members and key stakeholders. From six months onward, the role is expected to develop independent value in a particular area or task and to lead work as the project develops.

The assigned AI Development Division has approximately 15 members, including secondees from collaborating organizations, and operates in small groups of three to five people. Team support includes mentor one-on-one meetings, biweekly meetings with a manager, biweekly knowledge-sharing seminars, and weekly company meetings. The team uses Scrum for dialogue-driven hypothesis testing. The working arrangement is a discretionary work system with remote work as the primary format. Office attendance, short- or long-term business travel of up to one month, and on-site support may be required.

Named tools and environments include Python, PyTorch, GitHub, Slack, Google Workspace, Notion, Docker, Linux, computing clusters, and Raspberry Pi. Models are developed using servers. Depending on the project, models may be optimized for vehicle ECUs and deployed to Raspberry Pi or Jetson devices. Examples of client work include optimizing face landmark and gaze detection for an automotive driver-monitoring system to run on an inexpensive CPU, developing the edge AI camera ai cast for monitoring in smart-city and autonomous-vehicle contexts, and combining response retrieval, speech recognition, speech synthesis, and automatic behavior generation for the multimodal agent “Saya.”