Generative AI Projects (2026)

Generative AI projects that go beyond chat: retrieval (RAG), agents, prompt injection defence, evaluation and AI-assisted work in real tools.

18 projects: 4 beginner, 9 intermediate, 5 advanced. Every step is checked as you go, an AI mentor helps when you are stuck, and you finish with a verifiable certificate.

Generative AI projects

  1. FinOps on AWS: Bedrock + Lambda Cost Insights (with ChatGPT Integration) — learn finops using aws lambda and bedrock (intermediate · $2)
  2. GenAI Automation with n8n: Daily Productivity Workflows — Build a set of everyday automations in n8n that use GenAI to save time: a daily briefing, smart inbox triage, and meeting-note cleanup plus many more. (beginner · $1)
  3. AI-Powered RAG Customer Support Agent — Build an end‑to‑end customer issue resolution agent that answers repetitive queries (password resets, refund tracking, order status, policy lookups) using RAG. (intermediate · $1)
  4. Fine-Tuning LLMs: Dataset Design, LoRA, QLoRA, and Production Deployment — A deep, hands-on course on fine-tuning LLMs: dataset design, SFT/DPO/RFT, OpenAI hosted fine-tuning, local LoRA/QLoRA, evals, and a full AWS EC2 GPU capstone project. (advanced · $3)
  5. Agent2Agent (A2A) Protocol Explained: Agent Discovery, Messaging, and Multi-Agent Gateways — Learn the Agent2Agent (A2A) protocol: AgentCard discovery, messages/tasks/artifacts, JSON-RPC, streaming, and security for building a multi-agent gateway. (advanced · $2)
  6. How LLMs Actually Work: Tokens, Embeddings, and Transformers Explained — Learn how large language models actually work under the hood — tokenization, embeddings, transformer attention, and the training-versus-inference split — through DevOps analogies and hands-on labs. (beginner · $1)
  7. Build MCP Servers with FastMCP: Tools, Resources, and Prompts for Any AI Client — Learn to build real MCP servers with FastMCP, exposing tools, resources, and prompts that work with Claude Desktop, Cursor, and any MCP-compatible AI client. (intermediate · $2)
  8. LLMOps for Production GenAI: Cost, Latency, and Release Gates — Learn to run GenAI like production infrastructure: resilient model gateways, cost/latency budgets, prompt caching, deployment gates, and incident response. (advanced · $1)
  9. Local LLMs with Ollama: Run Production-Grade AI Models on Your Own Machine — Run production-grade open-weight language models locally with Ollama - no API key, no cloud costs, and no data leaving your machine. (beginner · $1)
  10. GenAI Evals and Red Teaming: Building Quality Gates for LLM Applications — Learn to design multi-dimensional LLM evals, red-team for prompt injection and unsafe agent behavior, and wire eval results into automated ship/block release gates. (intermediate · $1)
  11. GenAI Observability: Traces, Metrics, and OpenTelemetry for LLM Apps — Learn to instrument GenAI applications with traces, metrics, and OpenTelemetry semantic conventions so you can debug, evaluate, and audit LLM systems in production. (intermediate · $1)
  12. GenAI Security: Threat-Modeling LLM Apps with the OWASP Top 10 — Learn to threat-model GenAI applications using the OWASP LLM Top 10, design least-privilege agent tools, and produce security review records for RAG, MCP, and agent workflows. (advanced · $1)
  13. Modern AI Agent Runtime: State, Guardrails, Handoffs, and Traces — Compare agent runtime patterns, design durable state and typed handoffs, and build a minimal agent trace contract ready for eval-based release gates. (advanced · $1)
  14. Retrieval-Augmented Generation (RAG) Tutorial: Chunking, Embeddings, and Grounded Answers — Learn to build a complete RAG system, from PDF ingestion through chunking, embeddings, hybrid retrieval, reranking, and grounded answers with citations. (intermediate · $2)
  15. Tool Calling and Function Calling for LLMs: Building the Model-Tool-Model Loop — Learn how to design tool schemas and build a safety-hardened, provider-portable model-tool-model loop so LLMs can request real data and actions instead of guessing. (intermediate · $1)
  16. AI Governance, Risk, and Compliance: Building a GenAI Risk Register — Learn to build a GenAI risk register and versioned change records using the NIST AI RMF, then connect them to evals, observability, and release gates. (intermediate · $1)
  17. Memory Management in Agentic AI: Short-Term, Long-Term, and Episodic Memory Explained — Learn how to design short-term, long-term/semantic, episodic, reflective, and procedural memory for AI agents, with governance controls that keep it safe and trustworthy. (intermediate · $2)
  18. Prompt Engineering for Reliable AI Systems: Zero-Shot, Few-Shot, and Prompt Injection Defense — Learn to write system prompts, calibrate model behavior with few-shot examples, force reliable step-by-step reasoning, extract clean structured output, and defend AI tools against prompt injection. (beginner · $1)

Practise generative AI more

Frequently asked questions

Do I need to install anything for these generative AI projects?

Most projects run in labs inside your browser (SQL, Excel, Python, Power BI, Git and more). Cloud projects say up front when they use your own AWS account or a simulator instead.

How is a Bartman project different from a tutorial?

You do the work yourself, step by step. Each step is checked automatically, and an AI mentor helps when you are stuck, so you finish with work you can show rather than a video you watched.

Do I get a certificate?

Yes. Finishing a project earns a certificate with its own ID that anyone can verify on bartman.ai.

What is Bartman?

Bartman (bartman.ai) is a learning platform where you build real projects with an AI mentor, practise in free browser labs, take mock tests and practice exams, and run job simulations. It was founded in 2025 and is based in Hyderabad, India.

More project topics