Lead Embedded AI Software Engineer

About Bosch Engineering Center Cluj Since its establishment in 2013, Bosch Engineering Center Cluj plays an essential role in the current transformation of the mobility sector. Through its extensive expertise in software, hardware & mechanical engineering, reliability engineering, as well as sales planning, the center contributes to the development of innovative products and services based on artificial intelligence (AI) applied in the fields of automated driving, electric and connected mobility. Thanks to the diverse professional and personal development opportunities, flexible working conditions, modern offices and laboratories in Cluj-Napoca, Jucu, Bucharest and Sibiu, we are one of the most desired employers among engineers and IT professionals in Romania. Our promise to our colleagues is rock-solid: we grow together, we enjoy our work, and inspire each other. Join us and see the difference. Work LikeABosch We are looking for a Lead Embedded AI Software Engineer to design and deploy AI‑enabled embedded systems around a multi‑vendor AI chiplet platform. In this role, you will bridge hardware, low‑level AI accelerators, and complex software stacks to enable sustainable, safety‑ and security‑aware AI pipelines in embedded automotive or industrial environments. Key responsibilities: Define the embedded AI software architecture for multi‑vendor AI chiplet platforms. Drives host‑side Application processing unit (CPU) orchestration and code optimizations. Design and implement efficient hardware‑software interfaces (HSI) between AI chiplets and host systems via UCIe‑based communication, drivers, and access‑management layers. Implement and optimize embedded AI inference pipelines using C++ and OpenVX, targeting performance, latency, and resource constraints. Develop and maintain safety‑ and security‑monitoring components for AI workloads (e.g., health checks, secure execution, anomaly detection). Consults on embedded deployment, including model partitioning, slicing, retraining (or fine‑tuning) strategies, and quantization‑aware deployment. Lead the initial deployment and tuning of AI models on embedded targets, including memory, bandwidth, and power‑aware partitioning. Strong proficiency in C++ and embedded systems programming, with experience in real‑time or safety‑critical environments. Deep expertise in OpenVX, OpenCL, and/or GPU compute frameworks (e.g., Vulkan), especially for computer vision and/or AI pipelines. Solid understanding of AI chiplets, heterogeneous acceleration, and high‑speed interconnects (e.g., UCIe) and host‑side drivers. Familiarity with AI accelerator toolchains and neural network models to AI chiplet mapping. Strong systems‑thinking and cross‑layer design skills (from sensor data acquisition and pre-processing to inference and post-processing). Strong collaboration skills with hardware architects, firmware, safety/security, and algorithm teams. Good written and verbal communication skills, with the ability to document architecture, APIs, and integration guidelines. LikeABosch Benefits: Your work-life balance is valuable to us, so we offer you: 25 Days of annual leave , because work-life balance is essential to us; Flexible working hours. Your health and well-being is important to us, so we have you covered: Medical subscription; Accident insurance; Sport activities and well-being initiatives. We believe in a Life-Long Learning approach, so you can take advantage of: Technical and soft skills trainings; Access to e-learning platforms; Local and Global career development programs; Opportunity to attend conferences; Bookster subscription. You will be part of a product company . You will contribute to developing end-to-end products that are delivered to Global OEMs. On top of these, we are offering: A monthly budget which can be used for several different services on an online benefit platform; Annual bonus; Lunch discounts. Ready to work LikeABosch ? We live by our values, have an open feedback culture and there is plenty of room for innovative ideas. All because a team is only as valuable as its members.