AI Architect Tracker

Five certifications. One path.

Check off steps as you go. Each card shows your pre-flight requirements, progress through the course, and a cheat card that unlocks when you finish.

0/5
complete
1
2
3
4
5
Start with Google AI Essentials
0 / 40 steps
01
Google AI Essentials
~8 hrs · Self-paced · Coursera · Free
Enrol on Coursera →
Not started
Pre-flight
  • Free Google account (gmail.com)
  • Free Coursera account (coursera.org)
  • ~8 hours of study time available
0 of 8 steps complete 0%
Certificate
Google AI Essentials Certificate

Issued by Google via Coursera. Completion-based — no separate exam. Shareable on LinkedIn.

Cheat card unlocked Google AI Essentials
AI typesNarrow AI (task-specific) · AGI (theoretical) · ML is a subset of AI · LLMs are a form of Gen AI
ML typesSupervised (labelled data) · Unsupervised (find patterns) · Reinforcement (reward signal)
PromptingBe specific · give context · specify output format · iterate · zero-shot vs few-shot examples
Responsible AIFairness · Reliability & safety · Privacy & security · Inclusiveness · Transparency · Accountability
Google AI stackGemini (multimodal LLM) · NotebookLM (research) · Vertex AI (ML platform) · Workspace AI
HallucinationModel confidently outputs false info — mitigate with output verification, grounding, RAG, citations
02
GitHub Foundations
~12 hrs · Self-study + PSI exam · $99 USD
GitHub Skills →
Not started
Pre-flight
  • GitHub account (github.com/join)
  • Git installed locally (git-scm.com)
  • $99 USD exam budget
  • ~12 hours of study time available
0 of 7 steps complete 0%
Certificate
GitHub Foundations Certification

Issued by GitHub (Microsoft). ~60 questions · 120 minutes · 68% passing threshold · valid 2 years.

Cheat card unlocked GitHub Foundations
Core gitinit · clone · add · commit -m · push · pull · branch · checkout · merge · stash · log · diff
GitHub objectsrepo · branch · commit · PR · issue · fork · release · gist · package · project · discussion
GitHub Flowbranch → commit → open PR → review & discuss → approve → merge → deploy
Actions YAMLon: [push, pull_request] → jobs: → runs-on: ubuntu-latest → steps: → uses:/run:
SecretsSettings → Secrets → reference as ${{ secrets.NAME }} in workflow YAML
Exam~60 Qs · 120 min · 68% pass · PSI online proctored · $99 USD · valid 2 years
03
Azure AI Fundamentals (AI-900)
~20 hrs · Microsoft Learn + Pearson VUE · $165 USD
Microsoft Learn →
Not started
Pre-flight
  • Free Microsoft account (account.microsoft.com)
  • Free Azure subscription optional — labs run on free tier
  • $165 USD exam budget
  • ~20 hours of study time available
0 of 8 steps complete 0%
Certificate
Microsoft Certified: Azure AI Fundamentals

Issued by Microsoft. ~50–60 questions · 60 minutes · 700/1000 to pass. Does not expire.

Cheat card unlocked Azure AI Fundamentals (AI-900)
AI workloadsMachine learning · Computer vision · NLP · Document intelligence · Knowledge mining · Generative AI
Azure MLWorkspace → compute → dataset → train (AutoML/designer/SDK) → evaluate → deploy endpoint → monitor
Azure AI ServicesVision (OCR, face, object) · Language (CLU, QA, sentiment) · Speech (STT/TTS) · Decision
Azure OpenAIGPT-4o · DALL·E 3 · Whisper · Embeddings — deployed via Azure portal, governed by Microsoft
Responsible AIFairness · Reliability/safety · Privacy/security · Inclusiveness · Transparency · Accountability
Exam~50–60 Qs · 60 min · 700/1000 pass · Pearson VUE · $165 USD · no expiry
04
AWS AI Practitioner (AIF-C01)
~25 hrs · AWS Skill Builder + Pearson VUE · $100 USD
AWS Certification →
Not started
Pre-flight
  • Free AWS account (aws.amazon.com)
  • Free AWS Skill Builder account (skillbuilder.aws)
  • $100 USD exam budget
  • ~25 hours of study time available
0 of 9 steps complete 0%
Certificate
AWS Certified AI Practitioner (AIF-C01)

Issued by Amazon Web Services. 65 questions · 90 minutes · 700/1000 to pass · valid 3 years.

Cheat card unlocked AWS AI Practitioner (AIF-C01)
ML pipelineCollect → clean/label → split → train → tune (HPO) → evaluate → deploy → monitor → retrain
Gen AI conceptsFoundation model · parameters · fine-tuning · RLHF · RAG · embeddings · tokens · inference · temperature
AWS AI servicesBedrock (FMs) · SageMaker (MLOps) · Rekognition (vision) · Comprehend (NLP) · Transcribe · Polly · Kendra
Amazon BedrockServerless FM access · Agents (reasoning loops) · Knowledge Bases (RAG) · Guardrails · Titan models
Responsible AIFairness · Explainability · Privacy · Safety · Veracity/robustness · Transparency · Controllability
Exam65 Qs (15 unscored) · 90 min · 700/1000 pass · Pearson VUE · $100 USD · valid 3 years
05
CCA-F — Claude Certified Architect: Foundations
~15 hrs · Self-study + Anthropic exam · TBC
Anthropic Learn →
Not started
Pre-flight
  • Claude.ai account (claude.ai)
  • Read the Anthropic model cards and usage policies (anthropic.com/policy)
  • ~15 hours of study time available
0 of 8 steps complete 0%
Certificate
CCA-F — Claude Certified Architect: Foundations

Issued by Anthropic. Validates applied knowledge of Claude models, API, prompting, tool use, and responsible deployment.

Cheat card unlocked CCA-F — Claude Certified Architect: Foundations
ModelsHaiku 4.5 (fast/cheap) · Sonnet 4.6 (balanced) · Opus 4.8 (most capable, complex reasoning)
API essentialsPOST /v1/messages · model · messages[{role,content}] · max_tokens (required) · system · temperature 0–1
Context window200K tokens · vision (images + PDFs inline) · streaming (stream:true) · prompt caching (cache_control)
PromptingXML tags <task><context><examples> · chain of thought · few-shot · role in system · clear output format
Tool usetools[] with name/description/input_schema → tool_use block → tool_result in next user turn
SafetyConstitutional AI · AUP compliance · human-in-the-loop for high-stakes · prompt injection awareness
Claude CodeCLI + MCP servers · slash commands (/review, /compact) · hooks · CLAUDE.md memory · agent tasks