Repository with projects, exercises, and hands-on material developed during the AI Solutions Architecture Master program.
The course builds the ability to design and orchestrate end-to-end AI solutions: from translating business objectives into technical architectures, to selecting technologies, integrating AI models, and leading development teams.
Program completed: 400 hours, 11 capstone projects, professional certification in AI Solutions Architecture issued 2026-07-28 (certificate).
| # | Module | Capstone | Status |
|---|---|---|---|
| 00 | Python Programming | Warehouse order monitoring system | delivered |
| 01 | Machine Learning | Defective-parts classification pipeline | delivered |
| 02 | Large Language Models | Hybrid RAG for a company knowledge base (ChromaDB + BM25) | delivered |
| 03 | Agentic AI | Two capstones: DigitServe (n8n) + GreenThumb (ReAct + RAG) | delivered |
| 04 | Business Case & AI PM | OmniRetail AI governance platform | delivered |
| 05 | Cloud for AI | EnergoGrid hybrid multi-cloud infrastructure | delivered |
| 06 | AI Service Deployment | LogiFast delivery-time service (Flask) | delivered |
| 07 | Data Governance & KM | NovaCura Pharma data governance platform | delivered |
| 08 | Solution Design & Architecture | RetailSight video analytics platform | delivered |
| 09 | Information & Architecture Security | UrbanSight security risk assessment | delivered |
| 10 | Governance, Ethics & Compliance | AI governance framework for a fintech | delivered |
Programming fundamentals applied to real-world data engineering and automation scenarios. Project: Order and warehouse monitoring system.
Complete ML pipelines, from data preparation to model evaluation, with a focus on industrial use cases. Project: Classification model for defective parts in manufacturing.
Transformer architectures, prompt engineering, fine-tuning, and RAG systems for enterprise applications. Project: RAG system for intelligent enterprise knowledge management.
Design of multi-agent systems, orchestration, and tool use for automating complex processes. Projects: DigitServe agent orchestration (n8n) and GreenThumb support agent (ReAct + RAG).
Feasibility analysis, product roadmaps, cost estimation, and lifecycle management of AI solutions. Project: OmniRetail AI governance platform.
Design of cloud-native infrastructures for AI workloads on AWS, GCP, and Azure. Project: EnergoGrid hybrid multi-cloud AI infrastructure.
Containerization, CI/CD, monitoring, and scalability of models in production. Project: Deployment of a predictive model for delivery time estimation.
Strategies for data management, quality, and lineage in enterprise AI systems. Project: Data Governance architecture for a pharmaceutical company.
Architectural patterns for scalable, resilient, and maintainable AI systems. Project: RetailSight video analytics platform.
Threat modeling, vulnerability analysis, and hardening of AI-based systems: model attacks (adversarial examples, data poisoning, model inversion), data security, classic threats, supply chain, Zero Trust architectures, NIS 2, AI forensics. Project: Risk and vulnerability analysis of an AI-based system.
Regulatory frameworks (AI Act, GDPR), bias, explainability, and algorithmic accountability. Project: AI governance framework for a fintech company.
AI_solutions_architect_course/
├── 000_certs/ # Module certificates (00-10) + final master certificate (.pdf)
├── 00_python_programming/ # Notes + PRJ (warehouse monitoring, stdlib) - delivered
├── 01_machine_learning/ # Notes + exercises + PRJ (defective parts classification, sklearn) - delivered
├── 02_large_language_models/ # Notes + exercises + PRJ (hybrid RAG, ChromaDB + BM25) - delivered
├── 03_agentic_ai/ # Notes + exercises + 2 PRJ (n8n DigitServe + GreenThumb) - both delivered
│ # GreenThumb = LangChain/LiteLLM ReAct + RAG + FastAPI notebook (2nd module-03 capstone)
├── 04_business_case_AIPM/ # Notes + exercises + PRJ (OmniRetail AI governance platform) - delivered
├── 05_AI_cloud_services/ # Notes + exercises + PRJ (EnergoGrid hybrid multi-cloud) - delivered
├── 06_AI_services_deployment/ # Notes + exercises + slides (local) + PRJ (LogiFast delivery-time service, Flask) - delivered
├── 07_data_governance_knowledge_management/ # Notes (9) + slides (local) + PRJ (NovaCura DG&KM: PDF + 8 diagrams + SQL/BigData/KG/RAG artifacts) - delivered
├── 08_solutions_architectures_design/ # Notes (11) + slides (local) + exercises (ETL, FastAPI+Docker, mini-RAG, draw.io) + PRJ (RetailSight video analytics platform) - delivered
├── 09_information_and_architecture_security/ # Notes (8) + slides (local) + exercises (adversarial noise, Llama Guard paper, Zero Trust engine, NIS 2 / Belmont texts) + PRJ (UrbanSight security risk assessment: PDF + 5 diagrams) - delivered
├── 10_governance_ethics_compliance/ # Notes (10: AI governance, ethics, bias, risk, privacy, AI Act, GPAI, conformity, policies) + PRJ (CrediPulse AI governance framework: PDF + 5 diagrams) - delivered
└── README.md
Note: slide decks (.pptx) and capstone Word sources (.docx) are kept local and gitignored, so they are not part of the published repository. Every capstone delivered from module 04 onward ships its report as .pdf inside the _PRJ_* folder; the earlier capstones (modules 00-03) are code deliverables (.py, .ipynb, n8n .json).