Skip to content

Latest commit

 

History

History
50 lines (40 loc) · 2.5 KB

ROADMAP.md

File metadata and controls

50 lines (40 loc) · 2.5 KB

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).

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

  • 24-week curriculum with weekly READMEs, notebooks, and use cases
  • Knowledge base (Ng framework, LLM core, agents, platforms, Databricks)
  • Skill guides: Claude Code, Cursor, DeepSeek Harness, OpenCode, OpenRouter, local stacks
  • Agents section: fundamentals, multi-agent, MCP, OpenClaw, Hermes
  • Cloud platforms: Azure AI Foundry, Google Vertex AI, AWS Bedrock
  • Databricks zero-to-hero module (Unity Catalog → Lakeflow → Mosaic AI → production)
  • Excel 24-week progress tracker with visual dashboard
  • Branding: Zorost Intelligence, zorost.com
  • 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)
  • 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