# AI Engineering Lab: Roadmap

AI engineering changes weekly. This program is built to evolve. Everything below is
deliberate room for future improvement; PRs are welcome (see
[.github/CONTRIBUTING.md](.github/CONTRIBUTING.md)).

## v1.0: current (24-week core program)

- [x] 24-week curriculum with weekly READMEs, notebooks, and use cases
- [x] Knowledge base (Ng framework, LLM core, agents, platforms, Databricks)
- [x] Skill guides: Claude Code, Cursor, DeepSeek Harness, OpenCode, OpenRouter, local stacks
- [x] Agents section: fundamentals, multi-agent, MCP, OpenClaw, Hermes
- [x] Cloud platforms: Azure AI Foundry, Google Vertex AI, AWS Bedrock
- [x] Databricks zero-to-hero module (Unity Catalog → Lakeflow → Mosaic AI → production)
- [x] Excel 24-week progress tracker with visual dashboard
- [x] Branding: Zorost Intelligence, zorost.com
- [x] Rich lesson standard, every week README is a 2,000 to 3,500-word lesson
  (problem, deep concepts, diagrams, worked examples, pitfalls, glossary, quiz)

## v1.1: planned

- [ ] Video walkthroughs for each week (YouTube + embedded links)
- [x] Weekly quizzes, a 10-question `quiz.md` per week with answer key
- [ ] Docker-based lab environment (devcontainer with Jupyter + Ollama)
- [ ] GitHub Actions CI that smoke-tests every notebook on CPU
- [ ] GitHub Discussions templates for questions and week-by-week help

## v1.2: candidate new modules

- [ ] **Week 25+: AgentOps & Observability deep dive**, LangSmith, Langfuse, OpenTelemetry
  for agents; cost dashboards; production tracing
- [ ] **Evaluation engineering module**, LLM-as-judge calibration, synthetic eval data
  generation, red-teaming, regression suites
- [ ] **Multimodal track**, vision-language models, document AI, speech pipelines
- [ ] **Additional cloud platforms**, Snowflake Cortex, Oracle AI, IBM watsonx
- [ ] **Edge & mobile AI**, on-device inference (MLX, ExecuTorch, llama.cpp on mobile)
- [ ] **AI security & red-teaming track**, OWASP LLM Top 10, guardrails engineering
- [ ] **Regulated-industry track**, audit trails, lineage, compliance patterns
  (extension of Zorost's aviation/pharma/gov work)
- [ ] **AI product & spec shaping track**, Ng's "shaping the build" in depth:
  MVP decisions, user research for AI, refused-tradeoff playbooks

## Ideas welcomed

- Translations (content is MIT-licensed; community translations welcome)
- Regional cloud editions (more regions, sovereign-cloud notes)
- A "Fast Track" 8-week variant for experienced engineers

---
© 2026 Zorost Intelligence LLC · https://zorost.com