Safety for Managers
1 day
Training
Safety knowledge your team can put into practice.
Engineering-oriented training grounded in real development challenges. Build technical capability, strengthen decisions and connect learning to your own system.
01 / Technical training
Safe AI Engineering Training
A practical engineering course on the safety of AI-enabled systems, from hazards and requirements to architecture, runtime behavior and evidence.
For the people making engineering decisions
Safety engineers; system engineers and architects; AI/ML engineers; verification and validation engineers; technical managers.
Delivery that fits your team
On-site at the customer, remote / online, or company-specific workshops tailored to your own system and project. The format and practical activities are agreed with your team.
Day 1From Traditional Safety to Safe AI
Understand what changes
- Why AI changes safety engineering
- Deterministic versus probabilistic behavior
- Specification challenges and data dependency
- Distribution shift, confidence and uncertainty
- Systematic faults versus AI performance limitations
Keep the system in view
- The AI component versus the complete system
- Hazards first, AI second
- Translating system safety needs into AI-related requirements
- Interactive exercises and engineering case studies
Practice: examine a system-level hazard and identify the AI-related assumptions and requirements that need to be made explicit.
Day 2Engineering AI Safety Properties
Properties and measurable behavior
- Reliability, robustness, predictability and explainability
- Safety KPIs, false positives and false negatives
- Confidence, calibration and Brier score where useful
- Robustness and temporal stability
- Scenario / operational design domain (ODD) coverage
- Out-of-distribution (OOD) performance
Safety-oriented data engineering
- Dataset gaps and representativeness
- Rare scenarios and synthetic data
- Data traceability and safety-oriented review
- Connecting data/model weaknesses to safety-relevant properties
Practice: review a dataset and a set of model metrics, then define useful safety KPIs and evidence gaps.
Day 3Architecture, Runtime Safety and Assurance
Architecture and runtime safety
- Safe AI architecture and independent checks
- Runtime monitors and confidence monitoring
- Plausibility checks, guardrails and fallback
- Degraded operation and OOD handling
- Temporal consistency monitoring
Verification and the safety argument
- Verification and validation planning
- Scenario-based and robustness testing
- Fault injection and simulation
- Statistical evidence
- Safety argumentation / Goal Structuring Notation (GSN)
- Linking metrics and evidence to safety claims
Practice: outline a runtime safety concept and connect its verification evidence to a reviewable safety claim.
Day 4Optional: Simulation / Physical AI Lab
Interactive simulation activities
- Human/robot distance and closing-speed monitoring
- Stopping distance and safety envelopes
- Physical consequence awareness
- Stability, support relationships and cascading effects
From scenarios to evidence
- Collaborative maintenance scenarios
- Hazardous-energy situations
- Simulation-based evidence generation
- Review of assumptions, limitations and findings
Hands-on activities are selected for the delivery format and the customer’s system. Example cases can span automotive / automated driving, medical devices, robotics and industrial / infrastructure systems.
Training outcomes
Build a connected engineering approach.
Leave with a clearer understanding of the methods and decisions needed to engineer AI-enabled safety-critical systems.
- Understand what relevant standards and regulations require, recommend or influence for the system
- Identify AI-specific safety challenges
- Derive safety-relevant properties and KPIs
- Assess data and model weaknesses
- Design runtime monitors and guardrails
- Define Safe AI architecture
- Plan verification and simulation
- Build evidence toward future assessment
Standards and regulation support the engineering approach.
Relevant frameworks may include ISO 26262, ISO 21448 / SOTIF, ISO 8800 (ISO/PAS 8800), IEC 61508 and the EU AI Act. We identify what applies to the product and distinguish requirements, recommendations and supporting engineering practices.
ISO/IEC 42001 is covered at overview level as an AI management-system standard. It does not replace product-specific safety engineering. AI Act applicability depends on the system, use and organizational role; no automatic compliance outcome is implied.
Framework context: ISO road-vehicle standards · ISO/IEC 42001 overview · European Commission AI Act overview
02 / Management training
Functional Safety for Managers
A short, executive-summary-style training on the safety decisions and responsibilities managers need to understand.
Management decisions that shape safety
- Safety obligations and organizational responsibilities
- Planning, effort estimation and resources
- Management decisions, reviews and escalation
- Typical project risks and development interfaces
Safety planning in action
Use an interactive case study to connect safety culture, quality, workflow design and review readiness. Explore how responsibilities, priorities and escalation affect an actual safety plan.
Tailor the session to the organization’s maturity, product context and next project milestone.
Develop your team’s safety capability.
Share the roles, project context and learning objectives. We’ll discuss an appropriate training format.
Plan your training
Upcoming trainings
Choose a session below to request registration. All sessions start on a Tuesday; Safe AI includes three core days and an optional Friday lab.