# GapGuide role → O*NET-SOC code mapping
# Verified against onetonline.org on 2026-04-19. See research/06-dataset-sourcing.md §1.
# Composite roles list multiple SOCs; the parser unions their tech skills and dedupes.
#
# `industry` is GapGuide's own field (not from O*NET) — used for filtering on the Target Role page.
# `description_override` is optional; if blank the parser uses Occupation Data.txt's description.

roles:
  - name: "Data Scientist"
    industry: "Data Science"
    socs: ["15-2051.00"]
    description_override: ""

  - name: "Data Analyst"
    industry: "Data Science"
    socs: ["15-2041.00"]
    description_override: "Analyze data to identify patterns, generate insights, and produce reports for business decisions. Pakistan/global market role; O*NET groups under Statisticians (15-2041.00) which lists Data Analyst as a sample title."

  - name: "Machine Learning Engineer"
    industry: "AI / ML"
    socs: ["15-2051.00", "15-1221.00"]
    description_override: "Build, train, and deploy machine learning models in production. Combine data science (Data Scientists 15-2051.00) with computer science research (Computer and Information Research Scientists 15-1221.00). No dedicated O*NET SOC."

  - name: "Backend Developer"
    industry: "Software Engineering"
    socs: ["15-1252.00"]
    description_override: ""

  - name: "Frontend Developer"
    industry: "Software Engineering"
    socs: ["15-1254.00"]
    description_override: ""

  - name: "Full Stack Developer"
    industry: "Software Engineering"
    socs: ["15-1252.00", "15-1254.00"]
    description_override: "Build both server-side (Software Developers 15-1252.00) and client-side (Web Developers 15-1254.00) of web applications. Composite role — O*NET has no single SOC for full-stack."

  - name: "Data Engineer"
    industry: "Data Science"
    socs: ["15-1243.01", "15-1243.00"]
    description_override: "Design, build, and maintain data pipelines, warehouses, and ETL systems. Composite of Data Warehousing Specialists (15-1243.01) and Database Architects (15-1243.00). No dedicated O*NET SOC."

  - name: "DevOps Engineer"
    industry: "Infrastructure"
    socs: ["15-1299.08"]
    description_override: "Bridge development and operations through CI/CD, infrastructure as code, container orchestration, and cloud platform management. O*NET maps loosely to Computer Systems Engineers/Architects (15-1299.08); heavy manual augmentation needed for modern DevOps tooling."

  - name: "AI Engineer (GenAI / LLM)"
    industry: "AI / ML"
    socs: ["15-1221.00"]
    description_override: "Build applications powered by large language models, retrieval-augmented generation (RAG), and vector search. O*NET predates GenAI tooling — heavy manual augmentation required for LangChain, LlamaIndex, OpenAI/Anthropic API, prompt engineering, vector databases."

  - name: "Business Intelligence Analyst"
    industry: "Data Science"
    socs: ["15-2051.01"]
    description_override: ""