{ "schemaVersion": 1, "reviewedAt": "2026-11-04", "methodology": "Editorial learning suggestions, practice projects, and priority mappings. Linked resources support further learning, not mandatory role requirements or employment guarantees.", "skills": [ { "slug": "ai-literacy", "name": "Understand AI’s capabilities & limits", "category": "Foundations", "level": "Foundation", "description": "Recognize where generative AI helps and where human judgment is needed.", "why": "Choosing the right kind of task is more useful than trying every new tool.", "objectives": [ "Distinguish generation from prediction", "Recognize uncertainty and missing context", "Identify tasks needing human ownership" ], "practice": "Choose five recurring work tasks. Explain which could use AI, which should stay human-led, and why.", "proof": "A task map with clear boundaries and a reason for each choice.", "prerequisites": "No technical prerequisites", "tasks": [ "manage-projects", "find-team-knowledge" ], "resource": { "provider": "Microsoft Learn", "label": "Introduction to generative AI and agents", "url": "https://learn.microsoft.com/en-us/training/modules/fundamentals-generative-ai/" }, "reviewedAt": "2026-11-04" }, { "slug": "prompting-and-context", "name": "Give AI clear briefs & context", "category": "Foundations", "level": "Foundation", "description": "Turn a vague request into a clear goal, useful context, constraints, and a reviewable output.", "why": "Better instructions make it easier to judge whether the result meets your actual need.", "objectives": [ "State the audience and outcome", "Provide relevant examples and context", "Revise against explicit acceptance criteria" ], "practice": "Write three versions of a prompt for the same work task. Compare the outputs against a checklist you define first.", "proof": "A reusable brief and an annotated comparison of the three outputs.", "prerequisites": "No technical prerequisites", "tasks": [ "write-and-edit", "create-visuals" ], "resource": { "provider": "Microsoft Learn", "label": "Work smarter with AI", "url": "https://learn.microsoft.com/en-us/training/paths/work-smarter-with-ai/" }, "reviewedAt": "2026-11-04" }, { "slug": "verify-ai-output", "name": "Verify AI outputs", "category": "Foundations", "level": "Foundation", "description": "Check facts, numbers, citations, and omissions before relying on an AI response.", "why": "A polished answer can still be incomplete or wrong. Your review is part of the work.", "objectives": [ "Trace claims to original evidence", "Recalculate important numbers", "Record uncertainty and corrections" ], "practice": "Audit an AI-generated summary against the source document. Flag unsupported claims and write a corrected version.", "proof": "A claim-by-claim review with source locations and corrections.", "prerequisites": "No technical prerequisites", "tasks": [ "research-evidence", "write-and-edit", "analyze-data" ], "resource": { "provider": "NIST", "label": "AI Risk Management Framework", "url": "https://www.nist.gov/itl/ai-risk-management-framework" }, "reviewedAt": "2026-11-04" }, { "slug": "safe-data-handling", "name": "Handle data safely with AI", "category": "Foundations", "level": "Foundation", "description": "Decide what information an AI tool may access and who may see its outputs.", "why": "Useful assistance should not require exposing confidential work or personal information unnecessarily.", "objectives": [ "Classify data before sharing", "Review access and retention settings", "Use synthetic examples for experiments" ], "practice": "Create a data-sharing checklist for one workflow and test it using fictional records.", "proof": "A documented permission boundary and a redacted example input.", "prerequisites": "No technical prerequisites", "tasks": [ "find-team-knowledge", "automate-workflows" ], "resource": { "provider": "OWASP", "label": "GenAI application security resources", "url": "https://owasp.org/projects/top-10-for-large-language-model-applications" }, "reviewedAt": "2026-11-04" }, { "slug": "ai-data-analysis", "name": "Analyze data with AI assistance", "category": "Data & decisions", "level": "Applied", "description": "Use AI to help explore data while checking definitions, calculations, and missing values.", "why": "You need to distinguish a real business signal from a convincing chart or explanation.", "objectives": [ "Check data quality and units", "Verify generated formulas or queries", "Explain uncertainty to decision-makers" ], "practice": "Analyze a small public dataset, reproduce one AI-assisted calculation manually, and explain one limitation.", "proof": "A reproducible analysis with verified calculations and source data.", "prerequisites": "Basic spreadsheet skills", "tasks": [ "analyze-data", "finance-accounting" ], "resource": { "provider": "Microsoft Learn", "label": "Get started with Microsoft data analytics", "url": "https://learn.microsoft.com/en-us/training/paths/data-analytics-microsoft/" }, "reviewedAt": "2026-11-04" }, { "slug": "workflow-automation", "name": "Build reliable AI workflows", "category": "Automation", "level": "Applied", "description": "Connect an AI step to a repeatable process with checks, approvals, and a recovery path.", "why": "The useful skill is designing a process that works when the model or integration fails.", "objectives": [ "Map triggers and inputs", "Validate outputs before actions", "Test failures and human handoffs" ], "practice": "Build a sandbox workflow that drafts an update from sample records, pauses for review, and logs failed inputs.", "proof": "A working demo with a failure case and an approval step.", "prerequisites": "Comfort with a visual workflow editor", "tasks": [ "automate-workflows", "manage-projects" ], "resource": { "provider": "n8n Academy", "label": "Workflows, integrations, and AI testing", "url": "https://learn.n8n.io/" }, "reviewedAt": "2026-11-04" }, { "slug": "ai-assisted-coding", "name": "Develop & review code with AI", "category": "Software engineering", "level": "Applied", "description": "Use a coding assistant to implement changes, then inspect and test the result.", "why": "Engineering judgment still determines whether generated code solves the right problem.", "objectives": [ "Give repository-specific context", "Inspect diffs and dependencies", "Test behavior beyond the generated happy path" ], "practice": "Implement a small feature in a practice repository. Review the diff and add a regression test that fails before the fix.", "proof": "A reviewed change with meaningful tests and an explanation of tradeoffs.", "prerequisites": "Basic programming and version control", "tasks": [ "write-and-test-code" ], "resource": { "provider": "GitHub Docs", "label": "Review output from Copilot", "url": "https://docs.github.com/en/copilot/how-tos/copilot-on-github/use-copilot-agents/review-copilot-output" }, "reviewedAt": "2026-11-04" }, { "slug": "ai-evaluation", "name": "Evaluate AI systems", "category": "Software engineering", "level": "Advanced", "description": "Define representative tests and measure the errors that matter for a specific workflow.", "why": "A good demo does not establish reliability on unfamiliar inputs.", "objectives": [ "Build a representative test set", "Choose metrics for the cost of errors", "Compare versions against a baseline" ], "practice": "Create a small labeled classification dataset. Compare two approaches and explain false positives and false negatives.", "proof": "An evaluation report showing failure examples and a justified acceptance threshold.", "prerequisites": "Basic statistics and test design", "tasks": [ "write-and-test-code", "cybersecurity", "analyze-data" ], "resource": { "provider": "Google for Developers", "label": "Accuracy, recall, and precision", "url": "https://developers.google.com/machine-learning/crash-course/classification/accuracy-precision-recall" }, "reviewedAt": "2026-11-04" }, { "slug": "ai-security", "name": "Secure AI applications & agents", "category": "Security", "level": "Advanced", "description": "Review how untrusted input, tools, and permissions can affect an AI application.", "why": "An assistant with access to business systems needs carefully bounded authority.", "objectives": [ "Identify prompt-injection boundaries", "Restrict tool permissions", "Test sensitive-data exposure in a sandbox" ], "practice": "Threat-model a fictional internal assistant. Test malicious document instructions with dummy data and record defenses.", "proof": "A threat model, safe test cases, and a least-privilege tool policy.", "prerequisites": "Application security fundamentals", "tasks": [ "cybersecurity", "write-and-test-code" ], "resource": { "provider": "OWASP", "label": "GenAI application security resources", "url": "https://owasp.org/projects/top-10-for-large-language-model-applications" }, "reviewedAt": "2026-11-04" }, { "slug": "knowledge-retrieval", "name": "Ground AI in trusted knowledge", "category": "Knowledge systems", "level": "Applied", "description": "Organize approved sources so an assistant can retrieve relevant information and cite it.", "why": "Relevant context and access controls matter as much as the model you choose.", "objectives": [ "Curate current source documents", "Check retrieval relevance and citations", "Respect document-level permissions" ], "practice": "Build a small knowledge assistant over public documents. Test answerable, ambiguous, and unanswerable questions.", "proof": "A source inventory and a tested set of grounded answers and refusals.", "prerequisites": "Document organization; coding optional for a no-code prototype", "tasks": [ "find-team-knowledge", "research-evidence" ], "resource": { "provider": "Microsoft Learn", "label": "Retrieval-augmented generation in AI Search", "url": "https://learn.microsoft.com/en-us/azure/search/retrieval-augmented-generation-overview" }, "reviewedAt": "2026-11-04" }, { "slug": "ai-use-cases", "name": "Choose valuable AI use cases", "category": "Business leadership", "level": "Applied", "description": "Select a focused business problem and define what a successful AI pilot would change.", "why": "A clear outcome helps distinguish an improvement from a novelty.", "objectives": [ "Map process costs and constraints", "Set a baseline and success measure", "Compare build, buy, and non-AI options" ], "practice": "Write a one-page pilot proposal for one workflow with a baseline, owner, cost estimate, and stop condition.", "proof": "A decision-ready pilot brief with measurable success criteria.", "prerequisites": "No technical prerequisites", "tasks": [ "manage-projects", "analyze-data" ], "resource": { "provider": "Microsoft Learn", "label": "Transform your business with AI", "url": "https://learn.microsoft.com/en-us/training/paths/transform-your-business-with-microsoft-ai/" }, "reviewedAt": "2026-11-04" }, { "slug": "ai-change-management", "name": "Lead AI adoption & team learning", "category": "Business leadership", "level": "Applied", "description": "Help a team redesign work, learn new habits, and review the effects of AI adoption.", "why": "Adoption depends on useful workflows and feedback, not just access to software.", "objectives": [ "Map changing responsibilities", "Plan hands-on practice and feedback", "Measure quality alongside time saved" ], "practice": "Design a small team pilot with a training exercise, feedback channel, and review meeting.", "proof": "A rollout plan that names owners, support needs, and quality checks.", "prerequisites": "No technical prerequisites", "tasks": [ "manage-projects", "recruit-talent" ], "resource": { "provider": "Microsoft Learn", "label": "Transform your business with AI", "url": "https://learn.microsoft.com/en-us/training/paths/transform-your-business-with-microsoft-ai/" }, "reviewedAt": "2026-11-04" }, { "slug": "ai-governance", "name": "Set AI ownership & oversight", "category": "Business leadership", "level": "Advanced", "description": "Define who approves AI use, reviews risks, and responds when a system causes problems.", "why": "Teams need clear accountability as AI becomes part of normal operations.", "objectives": [ "Assign decision owners", "Document intended use and limits", "Set review and incident procedures" ], "practice": "Draft a lightweight use-case register for three fictional AI projects, including owners and review triggers.", "proof": "An ownership matrix and an escalation procedure the team can follow.", "prerequisites": "No technical prerequisites", "tasks": [ "manage-projects", "cybersecurity" ], "resource": { "provider": "NIST", "label": "AI Risk Management Framework", "url": "https://www.nist.gov/itl/ai-risk-management-framework" }, "reviewedAt": "2026-11-04" }, { "slug": "sales-research", "name": "Research accounts & personalize outreach", "category": "Sales & customer work", "level": "Applied", "description": "Use public evidence to prepare account briefs and useful, accurate sales messages.", "why": "Personalization works best when it reflects a real customer need rather than invented familiarity.", "objectives": [ "Separate evidence from assumptions", "Draft relevant questions and messages", "Keep CRM notes traceable" ], "practice": "Prepare a brief for a fictional prospect using public company information, then write an outreach draft with checked claims.", "proof": "A source-linked brief and a reviewed message matched to a stated need.", "prerequisites": "No technical prerequisites", "tasks": [ "sales-outreach", "write-and-edit" ], "resource": { "provider": "Microsoft Learn", "label": "Work smarter with AI", "url": "https://learn.microsoft.com/en-us/training/paths/work-smarter-with-ai/" }, "reviewedAt": "2026-11-04" }, { "slug": "service-agent-design", "name": "Design service agents & handoffs", "category": "Sales & customer work", "level": "Applied", "description": "Create support flows that use approved knowledge and hand over when they cannot help.", "why": "Service quality includes knowing when an automated answer is insufficient.", "objectives": [ "Define the agent’s scope", "Test knowledge-based answers", "Design escalation for unresolved requests" ], "practice": "Create a sample FAQ assistant and test ordinary questions, missing information, and a request requiring a human.", "proof": "A tested conversation set with a working human handoff.", "prerequisites": "No technical prerequisites", "tasks": [ "customer-support", "automate-workflows" ], "resource": { "provider": "Microsoft Learn", "label": "Test an agent’s knowledge sources", "url": "https://learn.microsoft.com/en-us/microsoft-copilot-studio/knowledge-test" }, "reviewedAt": "2026-11-04" }, { "slug": "ai-content-production", "name": "Create & review AI content", "category": "Creative work", "level": "Applied", "description": "Use AI for drafts and visual concepts while retaining editorial and brand control.", "why": "Output volume is less useful than content that is accurate, distinctive, and appropriate for its audience.", "objectives": [ "Write a creative brief", "Review claims and brand fit", "Check provenance and usage terms" ], "practice": "Produce three campaign concepts for a fictional product. Select one and document factual, accessibility, and rights checks.", "proof": "A finished asset with the brief, revisions, and review notes.", "prerequisites": "No technical prerequisites", "tasks": [ "create-visuals", "write-and-edit", "create-video-audio" ], "resource": { "provider": "Microsoft Learn", "label": "Work smarter with AI", "url": "https://learn.microsoft.com/en-us/training/paths/work-smarter-with-ai/" }, "reviewedAt": "2026-11-04" }, { "slug": "forecast-validation", "name": "Validate forecasts & planning scenarios", "category": "Operations", "level": "Applied", "description": "Review demand forecasts and distinguish model output from planning assumptions.", "why": "A forecast is useful only when you understand its errors and operational consequences.", "objectives": [ "Compare with a simple baseline", "Check seasonality and unusual events", "Explain scenario assumptions" ], "practice": "Use historical sample demand data to compare a forecast with a naive baseline and describe the biggest planning risk.", "proof": "A backtest and a scenario note explaining inventory or capacity implications.", "prerequisites": "Basic spreadsheets and forecasting concepts", "tasks": [ "plan-supply-chain", "analyze-data" ], "resource": { "provider": "Microsoft Learn", "label": "Analyze demand plans with Copilot", "url": "https://learn.microsoft.com/en-us/dynamics365/supply-chain/demand-planning/demand-planning-copilot" }, "reviewedAt": "2026-11-04" }, { "slug": "industrial-ai-validation", "name": "Evaluate industrial AI pilots", "category": "Operations", "level": "Advanced", "description": "Assess inspection or maintenance models against the conditions in which they will operate.", "why": "Performance on familiar examples may not hold across equipment, shifts, or sites.", "objectives": [ "Check representative operating data", "Identify costly missed detections", "Define supervised pilot boundaries" ], "practice": "Design a defect-detection pilot using sample images. Separate test conditions and document how errors would be reviewed.", "proof": "A pilot protocol with error criteria and a human review plan.", "prerequisites": "Domain process knowledge and basic statistics", "tasks": [ "analyze-data", "design-engineering" ], "resource": { "provider": "Google for Developers", "label": "Understanding overfitting", "url": "https://developers.google.com/machine-learning/crash-course/overfitting/overfitting" }, "reviewedAt": "2026-11-04" }, { "slug": "generative-design-review", "name": "Review generative engineering designs", "category": "Engineering", "level": "Advanced", "description": "Translate engineering requirements into constraints and review generated alternatives.", "why": "Generated geometry must still satisfy real design and manufacturing requirements.", "objectives": [ "Specify loads and constraints", "Compare design alternatives", "Document validation before production" ], "practice": "Set up a non-production design study and compare two outcomes against documented requirements.", "proof": "A design review that records constraints, tradeoffs, and remaining validation.", "prerequisites": "Engineering fundamentals and CAD experience", "tasks": [ "design-engineering" ], "resource": { "provider": "Autodesk", "label": "Generative design criteria", "url": "https://help.autodesk.com/cloudhelp/ENU/Fusion-GenerativeDesign/files/GD-DESIGN-CRITERIA.htm" }, "reviewedAt": "2026-11-04" }, { "slug": "research-literacy", "name": "Research with sources & uncertainty", "category": "Research", "level": "Applied", "description": "Use AI to organize evidence while inspecting the original material and its limitations.", "why": "A summary should help you examine evidence rather than replace that examination.", "objectives": [ "Track source provenance", "Separate findings from interpretation", "Identify conflicting or missing evidence" ], "practice": "Build an evidence table from three public documents. Compare an AI summary with the original claims.", "proof": "A source-linked synthesis with limitations and unresolved questions.", "prerequisites": "No technical prerequisites", "tasks": [ "research-evidence", "find-team-knowledge" ], "resource": { "provider": "Google", "label": "Gemini Notebook plans and source limits", "url": "https://support.google.com/gemininotebook/answer/16213268" }, "reviewedAt": "2026-11-04" }, { "slug": "ai-learning-design", "name": "Design learning with AI", "category": "Education", "level": "Applied", "description": "Use AI to prepare learning activities while preserving teacher judgment and learner agency.", "why": "A useful lesson supports understanding, not merely faster content generation.", "objectives": [ "Align activities to learning goals", "Review age and context suitability", "Design assessment that reveals understanding" ], "practice": "Adapt one lesson with an AI-assisted activity and a follow-up task that requires the learner’s own explanation.", "proof": "A lesson plan with learning goals, review notes, and an assessment rubric.", "prerequisites": "No technical prerequisites", "tasks": [ "lesson-planning", "create-visuals" ], "resource": { "provider": "UNESCO", "label": "AI competency framework for teachers", "url": "https://www.unesco.org/en/articles/ai-competency-framework-teachers" }, "reviewedAt": "2026-11-04" }, { "slug": "document-audit", "name": "Review AI-extracted documents", "category": "Professional practice", "level": "Applied", "description": "Check extracted fields and summaries against the original document and retain a review trail.", "why": "Small omissions in structured outputs can change downstream decisions.", "objectives": [ "Define an extraction schema", "Check every critical field", "Record source locations and reviewer decisions" ], "practice": "Extract dates and obligations from a fictional agreement or administrative form and verify them against the original.", "proof": "A reviewed extraction table with source references and unresolved items.", "prerequisites": "Familiarity with the relevant document type", "tasks": [ "legal-research", "clinical-documentation", "finance-accounting" ], "resource": { "provider": "NIST", "label": "AI Risk Management Framework", "url": "https://www.nist.gov/itl/ai-risk-management-framework" }, "reviewedAt": "2026-11-04" }, { "slug": "fair-ai-hiring", "name": "Assess fairness in AI-assisted hiring", "category": "People operations", "level": "Applied", "description": "Examine whether recruitment workflows rely on appropriate evidence and meaningful human review.", "why": "Automated recommendations need scrutiny before they influence opportunities for people.", "objectives": [ "Recognize biased data and proxies", "Compare outcomes across relevant groups", "Document human accountability" ], "practice": "Audit a fictional screening workflow using synthetic records. Identify questionable signals and redesign the review step.", "proof": "An audit of the workflow with fairness questions and an accountable review process.", "prerequisites": "Recruitment process knowledge; this is not legal compliance training", "tasks": [ "recruit-talent" ], "resource": { "provider": "Google for Developers", "label": "Machine learning fairness", "url": "https://developers.google.com/machine-learning/crash-course/fairness/" }, "reviewedAt": "2026-11-04" }, { "slug": "geospatial-ai-literacy", "name": "Interpret geospatial data for AI", "category": "Spatial & environmental work", "level": "Applied", "description": "Understand imagery, location, and time when preparing or reviewing spatial analysis.", "why": "Field conditions and data resolution determine what a map can reasonably support.", "objectives": [ "Distinguish raster and vector data", "Filter imagery by time and location", "Record resolution and coverage limits" ], "practice": "Compare public imagery for one area over two dates and explain what cannot be inferred reliably.", "proof": "An annotated map with provenance, time coverage, and limitations.", "prerequisites": "Basic mapping concepts; JavaScript useful for Earth Engine", "tasks": [ "analyze-data", "research-evidence" ], "resource": { "provider": "Google for Developers", "label": "Get started with Earth Engine", "url": "https://developers.google.com/earth-engine/guides/getstarted" }, "reviewedAt": "2026-11-04" }, { "slug": "sql-with-ai", "name": "Write & review SQL with AI", "category": "Databases & data", "level": "Applied", "description": "Draft SQL with an assistant, then check it against the real schema and known results.", "why": "Generated SQL can run without errors and still return the wrong answer.", "objectives": [ "Give the assistant the schema and business definitions", "Explain each join and filter before trusting a query", "Compare results with a figure you already know" ], "practice": "Ask an assistant for a monthly revenue query on a sample database. Then check row counts, null handling and duplicate joins against a hand-calculated total.", "proof": "A reviewed query with a note of what was checked and one error that was corrected.", "prerequisites": "Basic SELECT, JOIN and GROUP BY", "tasks": [ "database-sql" ], "resource": { "provider": "PostgreSQL Documentation", "label": "Queries", "url": "https://www.postgresql.org/docs/current/queries.html" }, "reviewedAt": "2026-11-04" }, { "slug": "query-tuning", "name": "Tune slow queries with AI", "category": "Databases & data", "level": "Advanced", "description": "Use AI to interpret execution plans, then prove each suggested change with measurements.", "why": "A suggested index can speed up one query while slowing writes and other queries.", "objectives": [ "Read an execution plan and find the costly step", "Test a rewrite or index on realistic data volumes", "Record the before and after timings and the write cost" ], "practice": "Capture EXPLAIN ANALYZE output for a slow query on a test database, ask an assistant for three options, and measure each one.", "proof": "A before-and-after plan comparison with the option you chose and why.", "prerequisites": "Comfortable writing SQL; access to a test database", "tasks": [ "database-tuning" ], "resource": { "provider": "PostgreSQL Documentation", "label": "Using EXPLAIN", "url": "https://www.postgresql.org/docs/current/using-explain.html" }, "reviewedAt": "2026-11-04" }, { "slug": "schema-design", "name": "Design schemas & review migrations", "category": "Databases & data", "level": "Applied", "description": "Use AI to draft data models and migrations, then review constraints, locks and rollback.", "why": "Schema mistakes are expensive to undo once production data depends on them.", "objectives": [ "Model entities, keys and constraints from real requirements", "Check a migration for locking and data loss risks", "Write and test a rollback or recovery path" ], "practice": "Draft a schema for a small ordering system with AI, then run its migration on a test copy with sample data and undo it.", "proof": "A reviewed schema and a migration tested in both directions.", "prerequisites": "Basic SQL and table design", "tasks": [ "data-modelling", "database-administration" ], "resource": { "provider": "PostgreSQL Documentation", "label": "Data Definition", "url": "https://www.postgresql.org/docs/current/ddl.html" }, "reviewedAt": "2026-11-04" }, { "slug": "safe-database-access", "name": "Give AI safe database access", "category": "Databases & data", "level": "Advanced", "description": "Connect AI tools to databases with limited permissions, and treat generated SQL as untrusted.", "why": "An assistant or agent with broad credentials can expose or destroy data through one bad statement.", "objectives": [ "Use read-only, least-privilege accounts for AI tools", "Keep production data away from tools that have not been approved", "Require parameterised queries and human approval for writes" ], "practice": "Create a read-only role on a test database, connect an assistant, and confirm that write attempts and restricted tables are refused.", "proof": "A tested access setup with the refused actions recorded.", "prerequisites": "Database users and permissions", "tasks": [ "database-administration", "database-sql" ], "resource": { "provider": "OWASP", "label": "SQL Injection Prevention Cheat Sheet", "url": "https://cheatsheetseries.owasp.org/cheatsheets/SQL_Injection_Prevention_Cheat_Sheet.html" }, "reviewedAt": "2026-11-04" }, { "slug": "database-recovery", "name": "Plan & test database recovery", "category": "Databases & data", "level": "Applied", "description": "Use AI to draft backup and recovery runbooks, then prove them with a real restore.", "why": "A backup only counts once a restore has worked within the time the business can accept.", "objectives": [ "Set recovery point and recovery time targets", "Draft a runbook and have it reviewed", "Restore to a separate environment and verify the data" ], "practice": "Back up a test database, delete a table, restore it by following an AI-drafted runbook, and correct whatever the runbook got wrong.", "proof": "A corrected runbook with the measured restore time.", "prerequisites": "Basic database administration", "tasks": [ "database-administration" ], "resource": { "provider": "PostgreSQL Documentation", "label": "Backup and Restore", "url": "https://www.postgresql.org/docs/current/backup.html" }, "reviewedAt": "2026-11-04" }, { "slug": "data-pipeline-testing", "name": "Test data pipelines with AI", "category": "Databases & data", "level": "Applied", "description": "Use AI to draft transformations and data tests, then confirm they catch real faults.", "why": "A broken pipeline can keep running and quietly feed wrong numbers into every report.", "objectives": [ "Define freshness, uniqueness and null checks for key tables", "Draft tests with AI and review what they miss", "Introduce a known fault and confirm a test fails" ], "practice": "Build one small transformation with tests on sample data, then add duplicate and missing rows and check that the tests catch them.", "proof": "A pipeline with tests, plus a record of the faults they caught.", "prerequisites": "SQL and basic pipeline concepts", "tasks": [ "data-pipelines" ], "resource": { "provider": "dbt Developer Hub", "label": "Add data tests to your DAG", "url": "https://docs.getdbt.com/docs/build/data-tests" }, "reviewedAt": "2026-11-04" }, { "slug": "hdl-review-with-ai", "name": "Review AI-generated RTL & HDL", "category": "Hardware & electronics", "level": "Applied", "description": "Use AI to draft Verilog, SystemVerilog or VHDL, then prove it with lint, simulation and synthesis.", "why": "Generated HDL often compiles but infers latches, misses resets, or breaks timing, and silicon mistakes are expensive.", "objectives": [ "Write a precise spec before prompting for RTL", "Lint, simulate and synthesise every generated block", "Spot latches, clock-domain crossings and reset bugs" ], "practice": "Solve ten HDL practice problems yourself, then ask an assistant for the same designs and compare results in simulation.", "proof": "A log of each generated module, the bugs simulation caught, and the fixes you made.", "prerequisites": "Digital logic basics and one HDL", "tasks": [ "chip-design", "fpga-design" ], "resource": { "provider": "HDLBits", "label": "Verilog practice problems with an online simulator", "url": "https://hdlbits.01xz.net/wiki/Main_Page" }, "reviewedAt": "2026-11-12" }, { "slug": "ai-assisted-verification", "name": "Use AI in design verification", "category": "Hardware & electronics", "level": "Advanced", "description": "Apply AI to testbenches, assertions, coverage and debug without weakening sign-off.", "why": "AI can speed up UVM boilerplate and failure triage, but coverage closure and sign-off stay an engineer’s responsibility.", "objectives": [ "Generate UVM components and assertions from a spec", "Use AI to triage regressions and waveforms", "Keep coverage and sign-off criteria independent of the AI" ], "practice": "Take a small open-source block, have AI draft its testbench and assertions, then measure the coverage and bugs it actually finds.", "proof": "A verification plan with coverage results and a note on what the AI missed.", "prerequisites": "SystemVerilog and basic UVM", "tasks": [ "chip-design" ], "resource": { "provider": "Verification Academy (Siemens)", "label": "UVM: Universal Verification Methodology", "url": "https://verificationacademy.com/topics/uvm-universal-verification-methodology/" }, "reviewedAt": "2026-11-12" }, { "slug": "ai-pcb-review", "name": "Check AI-generated schematics & PCB layouts", "category": "Hardware & electronics", "level": "Applied", "description": "Use AI for component selection, schematics and routing, then verify against datasheets, ERC and DRC.", "why": "AI tools can pick the wrong footprint, misread a pinout, or route a board that fails EMC, and a respin costs weeks.", "objectives": [ "Trace every AI-chosen part back to its datasheet", "Run ERC, DRC and power-budget checks", "Review high-speed, power and thermal choices by hand" ], "practice": "Design a small sensor board with AI help in KiCad, then check every pinout, footprint and rail against the datasheets.", "proof": "A design-review checklist showing each AI suggestion, how you checked it, and what you changed.", "prerequisites": "Basic circuit analysis", "tasks": [ "pcb-electronics" ], "resource": { "provider": "KiCad", "label": "Getting started in KiCad", "url": "https://docs.kicad.org/9.0/en/getting_started_in_kicad/getting_started_in_kicad.html" }, "reviewedAt": "2026-11-12" }, { "slug": "ai-firmware-on-hardware", "name": "Test AI-written firmware on real hardware", "category": "Hardware & electronics", "level": "Applied", "description": "Use AI to write and debug embedded C/C++, then confirm it on target hardware and under real timing.", "why": "Firmware that passes on a laptop can fail on the device: wrong registers, race conditions, stack overflow, or power issues.", "objectives": [ "Give the assistant the exact MCU, SDK and datasheet context", "Test on target with a debugger and logic analyser", "Check memory, timing and failure modes before shipping" ], "practice": "Have an assistant write a driver for a sensor on a dev board, then debug it on the board and record every correction.", "proof": "Working firmware with a test log from real hardware and a list of AI errors you caught.", "prerequisites": "C programming and one microcontroller", "tasks": [ "embedded-firmware" ], "resource": { "provider": "Zephyr Project", "label": "Zephyr RTOS getting started guide", "url": "https://docs.zephyrproject.org/latest/develop/getting_started/index.html" }, "reviewedAt": "2026-11-12" }, { "slug": "edge-ai-deployment", "name": "Deploy and check ML on embedded devices", "category": "Hardware & electronics", "level": "Advanced", "description": "Train, compress and deploy small models to microcontrollers, and measure accuracy, latency and power on the device.", "why": "Edge models lose accuracy when quantised and behave differently on real sensor data than in training.", "objectives": [ "Collect and label representative sensor data", "Quantise a model and measure on-device accuracy and latency", "Plan monitoring and updates in the field" ], "practice": "Build a keyword-spotting or vibration model, deploy it to a dev board, and compare on-device results with the training metrics.", "proof": "A short report with on-device accuracy, latency, memory use and known failure cases.", "prerequisites": "Basic Python and embedded development", "tasks": [ "embedded-firmware", "statistics-modelling" ], "resource": { "provider": "Edge Impulse", "label": "Edge Impulse documentation", "url": "https://docs.edgeimpulse.com/" }, "reviewedAt": "2026-11-12" }, { "slug": "safe-instrument-automation", "name": "Let AI drive instruments & controls safely", "category": "Hardware & electronics", "level": "Applied", "description": "Script lab instruments, test benches and plant controls with AI help, with limits, read-only modes and human approval.", "why": "An AI that sets the wrong voltage or writes the wrong PLC tag can damage equipment or hurt someone.", "objectives": [ "Automate instruments with SCPI and VISA", "Set hard limits, simulation and read-only modes first", "Keep a human approval step before any write to live equipment" ], "practice": "Write a measurement script with AI help against a simulated instrument, add limits and logging, then run it on real hardware under supervision.", "proof": "A tested script with its safety limits, a dry-run log and a sign-off record.", "prerequisites": "Basic Python and lab safety training", "tasks": [ "test-instrumentation", "plc-automation" ], "resource": { "provider": "PyVISA", "label": "PyVISA documentation: control instruments from Python", "url": "https://pyvisa.readthedocs.io/en/latest/" }, "reviewedAt": "2026-11-12" } ] }