Course Objective & Scope
AI projects do not behave like classic software projects: outcomes are uncertain, data quality drives the schedule and models must be monitored after go-live. This course teaches managers to break that uncertainty into manageable stages, make the right go/no-go gate decisions and set realistic expectations with stakeholders.
A CRISP-DM based lifecycle, PoC-to-production criteria, evaluation metrics and a RACI structure are applied to a sample project. Participants draft a project initiation document and a risk register for one of their own projects.
Who Should Attend?
- Managers sponsoring AI or data projects
- Project, programme and PMO managers
- Product owners and business analysts
- Procurement and IT managers working with AI vendors
Key Takeaways
- Split an AI project into PoC, pilot and production stages with gate criteria
- Translate business goals into measurable success metrics and acceptance criteria
- Build a risk register covering data, model and operational risks
- Design a RACI for data scientist, ML engineer and business owner roles
- Challenge vendor proposals on technical and commercial grounds
Curriculum
What Makes AI Projects Different?
- Experimentation, data dependency and probabilistic outputs
- Lessons from failed AI projects
CRISP-DM and the AI Project Lifecycle
- Business understanding, data understanding, preparation, modelling, evaluation, deployment
- PoC, pilot and production gate decisions
Success Criteria and Evaluation
- Linking business KPIs to model metrics
- Acceptance tests, human evaluation and eval sets
Team, Roles and Vendor Management
- Data scientist, ML engineer, product owner and domain expert roles
- Vendor selection, contracts and data rights
Risk, Compliance and Post-Production Management
- Data, model, security and regulatory risks
- Model monitoring, drift and continuous improvement responsibilities
- Exercise: project initiation document and risk register
Other Courses in This Category
-
What managers need to know about MLOps and LLMOps to move models from pilot to reliable service: lifecycle, monitoring and drift, evals, cost, platform options (MLflow, Azure AI Foundry, Bedrock, Vertex AI) and team structure.
- Live Course
- 5 modules
-
For business unit and strategy leaders: discovering AI opportunities across the value chain, scoring them on value, feasibility and risk, and building a balanced use case portfolio with prioritised use case cards.
- Live Course
- 5 modules
-
For finance, strategy and business leaders: measuring financial and operational returns of AI investments, covering TCO, token and infrastructure costs, productivity-to-P&L conversion, baselines and an ROI business case template.
- Live Course
- 5 modules
-
For executives, architects and procurement leaders: evaluating in-house build, SaaS buy and strategic partner options for AI solutions across competitive differentiation, TCO, speed to market and vendor lock-in.
- Live Course
- 5 modules
-
For strategy, PMO and transformation leaders: turning a use case portfolio into an executable three-horizon (90-day, 6-month, 18-month) enterprise AI roadmap aligned with data, platform and organisational dependencies.
- Live Course
- 4 modules
-
For IT, security, compliance and technology leaders: architectural, regulatory (KVKK, EU AI Act) and TCO trade-offs between cloud LLM APIs, managed enterprise cloud (Bedrock, Azure AI Foundry, Vertex AI) and self-hosted open-weight models.
- Live Course
- 5 modules
Get in touch
