This repository is a GitHub-ready portfolio plan for breaking into AI, AI automation, and AI app development. It starts with beginner projects that prove foundations, then moves into intermediate projects that look closer to real AI engineering work: RAG, agents, workflow automation, evals, deployment, and full-stack product thinking.
Build the projects in order. Projects 1-10 are beginner-friendly and should help you learn fast while producing visible GitHub proof. Projects 11-20 are intermediate and should become the strongest pinned repositories on your profile.
For every project, publish:
- A working demo or screen recording
- A clear README with problem, solution, architecture, setup, and screenshots
- Example inputs and outputs
- Tests or evaluation results
- A short "What I learned" section
- GitHub topics such as
ai,llm,rag,agents,automation,fastapi,nextjs, ormlops
Use one repo per serious project. Keep this repository as the master roadmap, then create separate repositories as you build each project. Pin your strongest 6 repositories on GitHub:
- Production RAG Knowledge Base
- Multi-Agent Research Assistant
- AI Workflow Automation Builder
- Customer Support AI Agent
- Codebase Documentation Agent
- AI Ops Evaluation Dashboard
- Prompt engineering and structured outputs
- API integration with LLMs and multimodal models
- Retrieval augmented generation with citations
- Agent design, tool use, orchestration, and human approval
- AI automation for email, calendar, support, documents, and workflows
- Full-stack app development with backend APIs and dashboards
- Evaluation, observability, cost tracking, and guardrails
- Practical deployment and documentation
If you can work 10-15 hours per week:
- Weeks 1-5: Finish projects 1-10 as small but polished apps
- Weeks 6-12: Build projects 11-16 with stronger architecture and demos
- Weeks 13-16: Build projects 17-20 and polish GitHub, portfolio site, resume, and case studies
- portfolio-roadmap.md has the full 20-project ladder.
- templates/project-readme-template.md is the README format to use for every GitHub repo.
- github-profile/README.md is a draft GitHub profile README.
- portfolio-positioning.md has profile and project positioning copy.
- projects contains a portfolio-ready brief for each project.
This roadmap is intentionally weighted toward implementation work because current AI portfolio signals favor people who can ship useful AI systems, not only explain models.
- OpenAI Agents SDK docs emphasize code-first agents, tool execution, orchestration, state, guardrails, human review, and observability: https://developers.openai.com/api/docs/guides/agents
- Hugging Face's AI Agents Course covers agent fundamentals, smolagents, LlamaIndex, LangGraph, agentic RAG, evaluation, and final projects: https://huggingface.co/learn/agents-course/en/unit0/introduction
- LangSmith evaluation docs highlight offline and online evaluation, RAG quality, agent trajectory evaluation, and CI integration: https://www.langchain.com/langsmith/evaluation
- LangChain's 2026 agent evaluation checklist recommends simple end-to-end evals, trace review, clear success criteria, and separating capability evals from regression evals: https://www.langchain.com/blog/agent-evaluation-readiness-checklist