About this role.
Sonatus is seeking a Senior AI Engineer to build production-grade agentic AI systems for software-defined vehicles. The role centers on multi-agent orchestration, tool calling, prompt management, LLM safety controls, and reliable cloud deployment at significant scale. The engineer will own the development lifecycle from prototyping through global deployment while collaborating with cross-functional stakeholders. Strong experience in LLMs, RAG, traditional machine learning, secure AI practices, and software engineering is required. The role is limited to candidates living in Poland, with remote work available within Poland.
Role DNA
A quick view of the complexity, pace, ownership and collaboration implied by the job description.
Job Complexity
5/5Pace & Pressure
5/5Autonomy Level
5/5Communication Load
4/5Salary analysis
Estimated compensation compared with the broader US market for similar roles.
Core skills
Skills and capabilities most closely associated with this opportunity.
Sample interview questions
I would define clear agent roles, bounded tool permissions, structured input and output schemas, and a planner-executor-verifier workflow. Each action would be validated against policy and operational constraints before execution, with durable tracing, idempotency, retries, and human escalation for high-risk operations.
I would combine data minimization, deterministic PII detection and redaction, input sanitization, policy classifiers such as jailbreak detectors, and strict separation between untrusted content and system instructions. Tool calls would use allowlists, least-privilege credentials, schema validation, and audit logs, while unsafe or uncertain requests would be blocked or routed for review.
I would store prompts as immutable, versioned artifacts with metadata for model compatibility, parameters, owners, evaluation results, and rollout status. Runtime routing would select prompt and model combinations through configuration or feature flags, and experiments would measure task success, latency, cost, safety failures, and user or operator feedback before promotion.
I would create an evaluation suite covering representative workflows, edge cases, adversarial prompts, tool failures, and safety scenarios. I would track task completion, factuality, policy compliance, tool-call correctness, latency, and cost, then supplement offline tests with staged production rollout, observability, and regression gates for every model, prompt, or tool change.
I separate experimentation from production paths and use prototypes to validate assumptions quickly. Once an approach proves valuable, I harden it with typed interfaces, tests, monitoring, rate limits, fallback behavior, deployment automation, and clear operational ownership so iteration does not compromise safety or service reliability.
At Sonatus, we’re driving the transformation to AI-enabled software-defined vehicles. Traditional automotive software methods can’t keep pace with consumer expectations shaped by the mobile industry—where features evolve rapidly, update seamlessly, and improve continuously. That’s why leading OEMs trust Sonatus to accelerate this shift. Our technology is already in production across more than 8 million vehicles on the road today and rapidly expanding.
Headquartered in Sunnyvale, CA, with 250+ employees worldwide, Sonatus combines the agility of a fast-growing company with the scale and impact of an established partner. Backed by strong funding and proven by global deployment, we’re solving some of the most interesting and complex challenges in the industry. Join us and help redefine what’s possible as we shape the future of mobility.
We are seeking a highly motivated AI Engineer to join our team and help us accelerate software innovations for next-generation software-defined vehicles. In this role, you will be responsible for the high-quality implementation of an Agentic Framework to support our AI applications. You will be responsible for writing the production-grade code that allows our AI applications to interact with millions of vehicles in real-time. We need a “builder” who thrives on the technical details of multi-agent orchestration, tool-calling optimisation, and LLM security. You will translate architectural designs into executable, high-performance systems, taking full ownership of the development lifecycle from local prototyping to global cloud deployment. This is a hands-on role for an engineer who prioritises execution, code quality, and system reliability for multi-agent systems, AI technologies and platforms.
This opportunity is for candidates who are already living in Poland. We do accommodate remote work within Poland.
Duties and Responsibilities
- Lead efforts to leverage existing AI models and frameworks to solve complex business challenges.
- Develop agentic orchestration by designing and implementing reason and act loops that allow AI agents to decompose complex goals into actionable sub-tasks.
- Design and implement a versioned prompt registry allowing for model-agnostic routing and A/B testing of system prompts
- Engineer the privacy shield for PII redaction and the hallucination verifier to audit AI-generated actions before execution
- Conduct the full cycle of data modeling and algorithm development, including modeling, training, tuning, validating, deploying, and maintaining services, (AI breadth).
- Strong domain expertise in the AI area including LLM, Time Series, RAG, fine-tune large models, traditional ML models, etc. (AI depth)
- Stay current with industry trends and advancements in data science and AI technologies. (State-of-the-Art)
- Perform data analysis and offer insights to inform business decisions across multiple domains.
- Adhere to data privacy and security protocols to uphold the confidentiality of sensitive information.
- Collaborate with cross-functional teams to understand requirements and translate them into effective AI and data science solutions.
- Document and communicate technical designs, processes, and best practices to stakeholders using visualizations and presentations.
- Take charge of projects, ensuring timely completion in a dynamic work environment.
Qualifications and Experience
- Master’s or PhD in Computer Science, Engineering, Mathematics, Applied Sciences, or a related field preferred. Bachelor’s degree required.
- Strong programming skills in languages such as Golang, Python, Java, or C++, with hands-on experience in relevant frameworks (e.g., TensorFlow, PyTorch, scikit-learn).
- Proven experience building multi-agent systems and frameworks with adaptive orchestration
- Deep understanding of Jailbreak Detection (e.g., LlamaGuard) and PII masking techniques to ensure safe LLM interactions in regulated environments
- In-depth knowledge and understanding of current machine learning algorithms, AI technologies, and platforms.
- Solid experience in data preprocessing, feature engineering, and model evaluation techniques.
- Familiarity with cloud platforms (e.g., AWS, Azure, Google Cloud) and containerization technologies (e.g., Docker, Kubernetes) is a plus.
- Strong knowledge of software development best practices, version control systems, and agile methodologies.
- Results-driven with a positive can-do attitude and excellent problem-solving skills.
- Exceptional verbal and written communication skills, with the ability to collaborate effectively with cross-functional teams.
- Experience in the automotive industry is highly desirable.
Annual salary information is not provided for this position. Explore salary ranges for similar roles in our Salary Directory ›
Upload your resume now
- To unlock remote work opportunities and be discovered by global employers.
This job listing has been manually reviewed by the Jobicy Trust & Safety Team for compliance with our posting guidelines, including verification of the company's legitimacy, accuracy of job details, clarity of remote work policy, and absence of misleading or fraudulent content.
