Curs Managementul proiectelor de implementare solutii de Inteligenta Artificiala – AI

Module 1 – Why AI Projects Are Different

AI projects challenge conventional project management assumptions. Unlike deterministic software, AI solutions are inherently probabilistic: they produce results in ranges of likelihood, not guaranteed correct outputs. Understanding these differences is essential for project managers.

• Deterministic vs. probabilistic systems and why this matters for planning
• The data dependency problem: AI projects succeed or fail based on data quality
• Model drift and the need for continuous monitoring post-deployment
• Why traditional PM methodologies (Waterfall, standard Agile) fall short for AI
• The role of the project manager in an AI team: what you own, what data scientists own
• Key vocabulary: model, training data, inference, overfitting, hallucination, foundation model

Module 2 – Introduction to CPMAI Methodology

A deep dive into the CPMAI framework—its origins, structure, and how it relates to CRISP-DM and Agile. Participants learn how to use the CPMAI Workbook as a living governance document across project iterations.

• Origins of CPMAI methodology: CRISP-DM and Agile
• CPMAI Methodology Artifacts
• Key Roles in AI Project Teams
  • Project Manager/AI Project Lead
  • Data Engineer
  • Business Analyst/Domain Expert
  • MLOps/DevOps Engineer
  • Executive Sponsor
  • Data Scientist
• Iterative nature of CPMAI methodology: when and how to revisit earlier phases
• Using the CPMAI methodology templates to document your AI projects
• CPMAI and Agile: how phases map to sprints and user stories

Module 3 – Phase I: Business Understanding

The first CPMAI phase establishes the foundation for the entire project. From a business perspective, this phase determines whether the AI project is worth pursuing and under what conditions it will be considered successful.

• Determining Business Fit for AI
  • Determine Business Objectives and Success Criteria
  • Cost-Benefit Analysis: project cost/time budget and expected ROI
• Mapping Business Needs to AI Patterns
• Translating Requirements Into Technical Strategy
• Simplifying Project Planning Through Pattern Identification
  • Recognition, Predictive Analytics and Decision Support, Conversational and Human Interaction, Goal-Driven Systems, Autonomous Systems, Anomaly Detection, Hyper-Personalization
• Defining ROI, Success Metrics, and Scope
  • ROI and Key Metrics: Establishing Clear Success Criteria
  • Scope Boundaries and MVPs: Enabling Rapid, Focused Delivery
• AI Go/No-Go Assessment
  • nine-factor assessment to determine whether an AI project should proceed: business feasibility, data feasibility, execution feasibility.

Module 4 – Phase II: Data Understanding

Cognitive projects are fundamentally dependent on having the right data in the right quality and format. Phase II tasks help project managers—without requiring data science expertise—to oversee data identification, collection, and quality assessment.

• Key Activities in PMI-CPMAI™ Phase II for Data Understanding
• Trustworthy AI Integration From the Start. Trustworthy AI frameworks (e.g., EU AI Act)
• Big Data Vs: Volume, Variety, Velocity, and Veracity
  • Evaluating Data Readiness for AI
  • Initial Data Collection Assessment
  • Data Analysis & Data Exploration
  • Assessing Data Quality. Data Quality Report: completeness, correctness, missing values, and required transformations
  • Assessment of Data Set Requirements for Training, Validation, and Test
  • Pretrained and Third-Party Model Usage: when to use foundation models, licensing, and cost considerations
• Conduct Go/No-Go assessment for next phase

Module 5 – Phase III: Data Preparation

Data preparation is the most time-consuming phase of any AI project. Project managers must understand what is being done, why, and what decisions are being made—even if they do not execute the work directly.

• Understanding Data Transformation Requirements
• Key Activities and Deliverables in Phase III
  • Data Selection: criteria for inclusion/exclusion of attributes and records
  • Data Cleansing: de-duplication, error correction, normalization, de-noising, anonymization
  • Data Enhancement: derived attributes, generated records, aggregation, third-party data merging
  • Data Labeling: identifying labeling needs, selecting labeling methods (internal, third-party, crowdsourcing), quality verification
• Establishing Data Governance and Compliance
• Transition to Model Development
• Building Repeatable Data Preparation Pipelines
  • Data Pipeline Documentation: training data pipeline vs. inference data pipeline
• Conduct Go/No-Go assessment for next phase

Module 6 – Phase IV: AI Model Development

Phase IV is where data scientists build and train models. Project managers must govern this phase without micromanaging technical decisions. The key is understanding enough to ask the right questions, evaluate tradeoffs, and make governance decisions.

• Obtaining the AI Model: Build-Versus-Buy-Versus-Integrate
• Algorithm Selection: supervised, unsupervised, reinforcement learning; how AI pattern guides this choice
• Ensemble Methods: combining multiple models to improve performance
• AutoML: automated model building tools, their benefits and limitations
  • Ai Model Training, Fine-Tuning, and Prompt Engineering
  • Custom Model Training and Development
  • Pretrained Model Fine-Tuning and Transfer Learning
  • Prompt Engineering and LLM Optimization
  • Hyperparameter Optimization: what hyperparameters are, why they matter, and how PMs oversee optimization cycles
• Performance Validation and Business Alignment
  • Model Test and Validation Design: train/test split, k-fold cross-validation, test design
• AI Trustworthiness in Development
• Conduct Go/No-Go assessment for next phase

Module 7 – Phase V: Model Evaluation

Model evaluation in CPMAI methodology is not just technical validation—it is a business governance activity. Project managers must interpret evaluation results and decide whether a model is ready to move forward, needs iteration, or requires fundamental rethinking.

• Strategic Objectives of Model Evaluation
  • Metric: What It Means for the PM
• Technical and Business Performance
• Understanding Model Performance Metrics: Accuracy, Precision, Recall (Sensitivity), F1 Score, False Positive Rate, False Negative Rate, etc
• Governance and MLOps Readiness
• Strategic Post-Deployment Planning and Life Cycle Management
• Conduct Go/No-Go assessment for next phase

Module 8 – Phase VI: Model Operationalization

The final CPMAI phase moves the approved model from the development environment into production use. Project managers must coordinate between the AI team and the broader IT and business stakeholders to operationalize the model responsibly.

• Primary Goals of Model Operationalization
• Key Activities in PMI-CPMAI™ Phase VI Model Operationalization
• Deployment Models and Strategies
  • Operationalization modes: batch processing, real-time API, microservice, streaming, on-premise vs. cloud
  • IT processes and change management required for model deployment
  • Model scaffolding: application development and integration work needed around the model
• Operationalizing GenAI Solutions
• Operationalizing AI Agents and Agentic AI Solutions
• Continuous Monitoring and Versioning
  • Monitoring and Maintenance Plan: data drift detection, performance degradation triggers, retraining schedules
• Project closure: Final Report and lessons learned

La finalizarea cu succes a acestui training, participanții vor fi capabili să realizeze:

• Planificarea și managementul proiectelor de implementare soluții cu Inteligenta Artificială AI de la inițiere până la operaționalizare, folosind metodologia CPMAI
• Înțelegerea cele 6 faze ale metodologiei CPMAI pentru implementarea soluțiilor AI
• Managementul activităților de pregătire a datelor fără a fi specializați în date
• Pregătirea pentru examenul de certificare PMI-CPMAI™  https://www.pmi.org/certifications/ai-project-management-cpmai

• Project manageri care conduc proiecte de implementare solutii cu Inteligenta Artificiala AI și Machine Learning ML
• Manageri de Program care supervizează portofolii ce includ inițiative cu Inteligenta Artificiala AI
• Analiști de Business Seniori și Manageri de Produs implicați în livrarea de soluții AI
• IT Manageri care achiziționează si implementează soluții de Inteligenta Artificiala (AI)
• Project Manageri care se pregătesc pentru examenul de certificare PMI-CPMAI™

Pentru a beneficia din plin de acest training sunt necesare urmatoarele cunostiinte preliminare:

• Implicare activă în management de proiect (minimum 1 an experiență)
• Familiaritate cu conceptele de bază Agile
• Nu este necesară experiență anterioară în data science sau machine learning
• Înțelegerea conceptelor a ceea ce reprezintă tehnologia inteligenta artificiala AI și Machine Learning ML
• Opțional: persoanele certificate PMP® sau CAPM® vor fi familiari cu aceasta abordare si cu conținutul trainingului

Nota: Acest training adresează managementul de proiect al inițiativelor de implementare soluții de Inteligenta Artificiala AI — nu pe construirea modelelor AI. Participanții nu au nevoie de experiență în programare, data science sau statistică.