Retour

Explorez tous les épisodes du podcast Cert & Chill

Plongez dans la liste complète des épisodes de Cert & Chill. Chaque épisode est catalogué accompagné de descriptions détaillées, ce qui facilite la recherche et l'exploration de sujets spécifiques. Suivez tous les épisodes de votre podcast préféré et ne manquez aucun contenu pertinent.

Rows per page:

1–50 of 53

TitreDateDurée
GAIL #11 · Exam Drills: Scenario Questions Across All Four Sections28 oct. 202600:19:01

Covers: Scenario questions across all four sections, product-to-scenario matching, all reflexes


Audio generated with AI from personal study notes; not official Google material. The official exam guide always wins on facts and weights: https://cloud.google.com/learn/certification/generative-ai-leader

GAIL #10 · Gen AI Strategy, SAIF and Responsible AI: Business First28 oct. 202600:15:25

Covers: Needs and resources before a project, top-down plus bottom-up strategy, measuring impact; SAIF, IAM, Security Command Center; responsible AI, privacy, bias, accountability, explainability


Audio generated with AI from personal study notes; not official Google material. The official exam guide always wins on facts and weights: https://cloud.google.com/learn/certification/generative-ai-leader

GAIL #9 · Prompting, RAG and Grounding: Company Facts vs World Facts28 oct. 202600:14:23

Covers: Zero-shot, one-shot, few-shot, role prompting, prompt chaining, chain-of-thought, ReAct, metaprompting; grounding with first-party, third-party and world data; RAG; temperature, top-p, token count, safety settings, output length


Audio generated with AI from personal study notes; not official Google material. The official exam guide always wins on facts and weights: https://cloud.google.com/learn/certification/generative-ai-leader

GAIL #8 · Hallucinations and Model Limits: Match the Fix28 oct. 202600:10:06

Covers: Data dependency, knowledge cutoff, bias, fairness, hallucinations, edge cases; grounding, RAG, prompt engineering, fine-tuning, HITL; versioning, drift monitoring, Feature Store, Pipelines


Audio generated with AI from personal study notes; not official Google material. The official exam guide always wins on facts and weights: https://cloud.google.com/learn/certification/generative-ai-leader

GAIL #7 · Agent Platform and AI Agents: Prototype vs Production28 oct. 202600:13:54

Covers: Agent Platform, Model Garden, Model Builder and AutoML, custom agents, agent components and types (deterministic, generative, hybrid; customer, employee, creative, data, code, security), tools (extensions, functions, data stores, plugins), pre-built AI APIs, Google AI Studio vs Agent Studio


Audio generated with AI from personal study notes; not official Google material. The official exam guide always wins on facts and weights: https://cloud.google.com/learn/certification/generative-ai-leader

GAIL #6 · Agent Search, Agent Assist and CX Insights: Who Serves Whom28 oct. 202600:17:06

Covers: Agent Search for websites and enterprise data; Google Search; the customer experience suite: CX Agent Studio (conversational agents), Agent Assist, CX Insights, Contact Center as a Service


Audio generated with AI from personal study notes; not official Google material. The official exam guide always wins on facts and weights: https://cloud.google.com/learn/certification/generative-ai-leader

GAIL #5 · Gemini, Gemini Enterprise or Workspace: Pick the Right One28 oct. 202600:13:24

Covers: Gemini app and Gems, Gemini Enterprise (enterprise search, assistant, custom agents, Gemini Notebook), Gemini in Google Workspace, Gemini for Google Cloud


Audio generated with AI from personal study notes; not official Google material. The official exam guide always wins on facts and weights: https://cloud.google.com/learn/certification/generative-ai-leader

GAIL #4 · Why Google Cloud: Open, Enterprise-Ready, Full-Stack28 oct. 202600:16:23

Covers: AI-first approach, enterprise-ready platform, open approach, integrated ecosystem, AI-optimized infrastructure (AI Hypercomputer, TPUs, GPUs), data control, low-code and no-code tools


Audio generated with AI from personal study notes; not official Google material. The official exam guide always wins on facts and weights: https://cloud.google.com/learn/certification/generative-ai-leader

GAIL #3 · Data, AI Layers and Google Models: Modality First28 oct. 202600:13:28

Covers: Data quality and accessibility, structured vs unstructured, labeled vs unlabeled; the five layers (infrastructure, models, platforms, agents, applications); Gemini, Gemma, Imagen, Veo; choosing a foundation model


Audio generated with AI from personal study notes; not official Google material. The official exam guide always wins on facts and weights: https://cloud.google.com/learn/certification/generative-ai-leader

GAIL #2 · Gen AI and ML Fundamentals: Labels Decide the Approach28 oct. 202600:15:34

Covers: AI, ML, deep learning, gen AI, foundation models, LLMs, multimodal and diffusion models; supervised, unsupervised and reinforcement learning; the ML lifecycle and its Google Cloud tools


Audio generated with AI from personal study notes; not official Google material. The official exam guide always wins on facts and weights: https://cloud.google.com/learn/certification/generative-ai-leader

GAIL #1 · How the Exam Works: Think Like a Leader, Not an Engineer28 oct. 202600:12:25

Covers: Who the exam is for, 4 sections and weights, 50-60 multiple choice questions in 90 minutes, pass or fail only, $99, valid 3 years, retakes, recent product renaming


Audio generated with AI from personal study notes; not official Google material. The official exam guide always wins on facts and weights: https://cloud.google.com/learn/certification/generative-ai-leader

AI-901 #9 · Domain drills20 oct. 202600:25:06

Covers: Questions across both domains, scenario-to-service matching, all reflexes


Audio generated with AI from personal study notes; not official Microsoft material. The official exam guide always wins on facts and weights: https://learn.microsoft.com/en-us/credentials/certifications/resources/study-guides/ai-901

AI-901 #8 · Information extraction with Content Understanding20 oct. 202600:22:39

Covers: Azure Content Understanding in Foundry Tools: documents and forms, images, audio and video; analyzers and schemas; structured output


Audio generated with AI from personal study notes; not official Microsoft material. The official exam guide always wins on facts and weights: https://learn.microsoft.com/en-us/credentials/certifications/resources/study-guides/ai-901

AI-901 #7 · Vision and image generation in Foundry20 oct. 202600:19:38

Covers: Multimodal models that read images; generative models that create images and video; lightweight vision apps


Audio generated with AI from personal study notes; not official Microsoft material. The official exam guide always wins on facts and weights: https://learn.microsoft.com/en-us/credentials/certifications/resources/study-guides/ai-901

AI-901 #6 · Text and speech in Foundry20 oct. 202600:23:54

Covers: Text analysis (keyword extraction, entity detection, sentiment, summarization) with Azure Language in Foundry Tools; speech to text and text to speech with Azure Speech; responding to spoken prompts with a multimodal model


Audio generated with AI from personal study notes; not official Microsoft material. The official exam guide always wins on facts and weights: https://learn.microsoft.com/en-us/credentials/certifications/resources/study-guides/ai-901

AI-901 #5 · Generative AI apps and agents in Foundry20 oct. 202600:26:45

Covers: System and user prompts; deploy a model and use it in the portal; chat client with the Foundry SDK; single-agent solution; client app for an agent


Audio generated with AI from personal study notes; not official Microsoft material. The official exam guide always wins on facts and weights: https://learn.microsoft.com/en-us/credentials/certifications/resources/study-guides/ai-901

AI-901 #4 · AI workloads20 oct. 202600:20:17

Covers: Scenarios for generative and agentic AI, text analysis, speech, computer vision and image generation, information extraction; techniques for each


Audio generated with AI from personal study notes; not official Microsoft material. The official exam guide always wins on facts and weights: https://learn.microsoft.com/en-us/credentials/certifications/resources/study-guides/ai-901

AI-901 #3 · How generative AI models work20 oct. 202600:21:51

Covers: How LLMs work; choosing a model by its capabilities; model deployment options and configuration parameters, including the Foundry deployment types


Audio generated with AI from personal study notes; not official Microsoft material. The official exam guide always wins on facts and weights: https://learn.microsoft.com/en-us/credentials/certifications/resources/study-guides/ai-901

AI-901 #2 · Responsible AI19 oct. 202600:18:10

Covers: Six principles: fairness, reliability and safety, privacy and security, inclusiveness, transparency, accountability; examples; content filters and the Discover, Protect, Govern approach


Audio generated with AI from personal study notes; not official Microsoft material. The official exam guide always wins on facts and weights: https://learn.microsoft.com/en-us/credentials/certifications/resources/study-guides/ai-901

AI-901 #1 · Strategy and how the exam works19 oct. 202600:12:33

Covers: AI-900 retired and AI-901 is the exam now; two domains (40-45% concepts, 55-60% implement with Microsoft Foundry); 700 of 1,000 scaled score; 45 minutes for Fundamentals exams; no penalty for guessing; retakes; Fundamentals certifications do not expire; Python and Azure familiarity expected


Audio generated with AI from personal study notes; not official Microsoft material. The official exam guide always wins on facts and weights: https://learn.microsoft.com/en-us/credentials/certifications/resources/study-guides/ai-901

AIF-C01 #9 · Domain drills16 oct. 202600:19:52

Covers: Questions across all five domains, ordering and matching practice, all reflexes


Audio generated with AI from personal study notes; not official AWS material. The official exam guide always wins on facts and weights: https://aws.amazon.com/certification/certified-ai-practitioner/

AIF-C01 #8 · Cross-domain: the AWS service map16 oct. 202600:26:05

Covers: Which in-scope service fits which scenario (ML, analytics, database, security, governance); managed service vs custom


Audio generated with AI from personal study notes; not official AWS material. The official exam guide always wins on facts and weights: https://aws.amazon.com/certification/certified-ai-practitioner/

AIF-C01 #7 · Domain 5: Security, Compliance, and Governance16 oct. 202600:24:46

Covers: 5.1 IAM, encryption, Macie, PrivateLink, shared responsibility, prompt injection, data leakage, grounding and hallucination detection; 5.2 Config, Inspector, Artifact, CloudTrail, Trusted Advisor, data governance, Security Scoping Matrix


Audio generated with AI from personal study notes; not official AWS material. The official exam guide always wins on facts and weights: https://aws.amazon.com/certification/certified-ai-practitioner/


AIF-C01 #6 · Domain 4: Guidelines for Responsible AI (14%)16 oct. 202600:23:57

Covers: 4.1 bias, fairness, safety, Guardrails, legal risks, dataset traits, bias vs variance; 4.2 transparency, explainability, Model Cards, human-centered design


Audio generated with AI from personal study notes; not official AWS material. The official exam guide always wins on facts and weights: https://aws.amazon.com/certification/certified-ai-practitioner/ (based on Exam Guide v1.1, April 2026; price, validity and requirements are on that page too).

AIF-C01 #5 · Domain 3b: Customizing and evaluating FMs16 oct. 202600:22:57

Covers: 3.1 customization cost tradeoffs and distillation; 3.3 training, fine-tuning, data preparation, RLHF; 3.4 evaluation methods and metrics, evaluating RAG and agents


Audio generated with AI from personal study notes; not official AWS material. The official exam guide always wins on facts and weights: https://aws.amazon.com/certification/certified-ai-practitioner/ (based on Exam Guide v1.1, April 2026; price, validity and requirements are on that page too).

AIF-C01 #4 · Domain 3a: Designing with FMs and prompting15 oct. 202600:23:24

Covers: 3.1 model selection criteria, inference parameters, RAG and Knowledge Bases, vector stores on AWS, agents; 3.2 prompt techniques, risks, Prompt Management


Audio generated with AI from personal study notes; not official AWS material. The official exam guide always wins on facts and weights: https://aws.amazon.com/certification/certified-ai-practitioner/ (based on Exam Guide v1.1, April 2026; price, validity and requirements are on that page too).

AIF-C01 #3 · Domain 2: Fundamentals of GenAI (24%)15 oct. 202600:22:28

Covers: 2.1 tokens, embeddings, chunking, FM lifecycle, token pricing, context engineering, agentic concepts and MCP; 2.2 strengths and limits; 2.3 AWS GenAI services and cost tradeoffs


Audio generated with AI from personal study notes; not official AWS material. The official exam guide always wins on facts and weights: https://aws.amazon.com/certification/certified-ai-practitioner/ (based on Exam Guide v1.1, April 2026; price, validity and requirements are on that page too).

AIF-C01 #2 · Domain 1: Fundamentals of AI and ML (20%)15 oct. 202600:22:25

Covers: 1.1 terms, learning types, data types, inference types; 1.2 use cases, when NOT to use AI, traditional ML vs FM, managed AI services; 1.3 pipeline, MLOps, metrics


Audio generated with AI from personal study notes; not official AWS material. The official exam guide always wins on facts and weights: https://aws.amazon.com/certification/certified-ai-practitioner/ (based on Exam Guide v1.1, April 2026; price, validity and requirements are on that page too).

AIF-C01 #1 · Strategy and how the exam works15 oct. 202600:13:37

Covers: Who it is for, 65 items (50 scored + 15 unscored), 90 min, 700/1000, compensatory scoring, four question types (multiple choice, multiple response, ordering, matching), five domains and weights, what is new in v1.1.


Audio generated with AI from personal study notes; not official AWS material. The official exam guide always wins on facts and weights: https://aws.amazon.com/certification/certified-ai-practitioner/ (based on Exam Guide v1.1, April 2026; price, validity and requirements are on that page too).

CCAO-F #9 · Domain Drills: All Seven Domains05 oct. 202600:20:50

The finale of the CCAO-F Bootcamp. One examiner, one candidate, and scenario questions across all seven domains, including "select TWO" items. Each answer is defended, and each wrong option is taken apart. Questions are tagged by domain so you can track your score per domain, and the episode closes with the reflexes that decide most questions.


Audio generated with AI from personal study notes; not official Anthropic material. The official exam guide always wins on facts and weights: https://www.anthropic.com/learn/certification

CCAO-F #8 · Troubleshooting: Diagnose Before You Retry05 oct. 202600:23:14

Domain 7 of the CCAO-F exam (10%): Troubleshooting and Optimization. A simple way to diagnose weak results: vague prompt, missing context, wrong model or feature, stale knowledge, a very long chat, or a limit no prompt can fix. How to adjust one thing at a time, and when to escalate technical work. Rule of the episode: diagnose before retrying.


Audio generated with AI from personal study notes; not official Anthropic material. The official exam guide always wins on facts and weights: https://www.anthropic.com/learn/certification



CCAO-F #7 · Governance and Responsible Use: Sensitive Data First05 oct. 202600:23:33

Domain 6 of the CCAO-F exam (15%): Governance, Risk, and Responsible Use. Appropriate and inappropriate use cases, handling sensitive and regulated data, following your organization's AI policy, and the ethics of AI use. Why anonymizing data before you upload it beats asking Claude not to retain it. This episode teaches the exam's reasoning and is not legal advice.


Audio generated with AI from personal study notes; not official Anthropic material. The official exam guide always wins on facts and weights: https://www.anthropic.com/learn/certification

CCAO-F #6 · Configuration and Knowledge: Projects Done Right05 oct. 202600:22:39

Domain 5 of the CCAO-F exam (12%): Configuration and Knowledge Management. What belongs in a Project's instructions and what belongs in its knowledge, how to curate and maintain uploaded files, how connectors like Google Drive and Gmail fit in, and why stale knowledge fails silently. Rule of the episode: stable context lives in the Project, and you keep it current.


Audio generated with AI from personal study notes; not official Anthropic material. The official exam guide always wins on facts and weights: https://www.anthropic.com/learn/certification

CCAO-F #5 · Workflow Integration: Redesign the Process, Not Just the Tool05 oct. 202600:21:40

Domain 4 of the CCAO-F exam (16%): Workflow Integration and Solution Design. How to use Claude to analyze requirements, plan and improve a process, when to augment a step and when to redesign the whole workflow, how to be honest with stakeholders about value and limits, and when to escalate technical work to Architects and Developers.


Audio generated with AI from personal study notes; not official Anthropic material. The official exam guide always wins on facts and weights: https://www.anthropic.com/learn/certification

CCAO-F #4 · Product and Model Selection: The Lightest Option That Works05 oct. 202600:21:53

Domain 3 of the CCAO-F exam (12%): Product and Model Selection. When to use a Project, Research, a plain chat or an artifact, how to match Haiku, Sonnet and Opus to the task, and what to do about long conversations and memory: when to restart, when to summarize, and when to save key decisions in a Project. Rule of the episode: use the lightest option that does the job.


Audio generated with AI from personal study notes; not official Anthropic material. The official exam guide always wins on facts and weights: https://www.anthropic.com/learn/certification

CCAO-F #3 · Output Evaluation: Verify Before You Share05 oct. 202600:15:31

Domain 2 of the CCAO-F exam (21%, the heaviest domain): Output Evaluation and Validation. Why confident answers are not accurate answers, which details are riskiest (citations, numbers, dates, names), how to fact-check, when a human must review the output, and how to adapt and format results for the audience. Rule of the episode: verify before you share.


Audio generated with AI from personal study notes; not official Anthropic material. The official exam guide always wins on facts and weights: https://www.anthropic.com/learn/certification

CCAO-F #2 · Prompting and Task Execution: Specific, Split, Iterate05 oct. 202600:21:18

Domain 1 of the CCAO-F exam (14%): Prompting and Task Execution. How to write prompts that give Claude an audience, purpose, format and constraints, how to split a big deliverable into steps, how to iterate with specific feedback instead of repeating the request, and how the approach changes between brainstorming, drafting and analysis.


Audio generated with AI from personal study notes; not official Anthropic material. The official exam guide always wins on facts and weights: https://www.anthropic.com/learn/certification

CCAO-F #1 · How the Exam Works: Format, Weights and Who It's For05 oct. 202600:20:53

How the Claude Certified Associate – Foundations (CCAO-F) exam works, and how to read its questions. We cover the format (60 items, 120 minutes, pass at a scaled 720 out of 1,000), the seven domains and their weights, who the exam is for and who it is not, retakes, renewal, and why the best answer is not always the first reasonable one. We also walk through the three official sample questions.


Audio generated with AI from personal study notes; not official Anthropic material. The official exam guide always wins on facts and weights: https://www.anthropic.com/learn/certification

CCAR-F #6 · Context and Reliability: Never Hide a Failure05 oct. 202600:48:39

Domain 5 (15%). Keeping critical facts alive in long sessions, escalation decisions, error propagation across multi-agent systems, context degradation in large codebase exploration, human review with confidence calibration, and preserving provenance when sources conflict. We run a fictional production incident as a post-mortem to find the root cause.


AI-generated hosts. Independent study aid, not affiliated with Anthropic. The official Exam Guide is the source of truth.

CCAR-F #4 · Claude Code Config: Match the Need to the Mechanism05 oct. 202600:58:39

Domain 3 (20%). The CLAUDE.md hierarchy, path-scoped rules in .claude/rules/, commands versus skills and their frontmatter, plan mode versus direct execution, iterative refinement, and running Claude Code in CI with -p, --output-format json and --json-schema. We contrast soft instructions with hard controls throughout.


AI-generated hosts. Independent study aid, not affiliated with Anthropic. The official Exam Guide is the source of truth.

CCAR-F #7 · Scenario Drills: All Six Exam Scenarios05 oct. 202600:48:30

The finale. We drill all six published scenarios (support resolution agent, code generation with Claude Code, multi-agent research, developer productivity, Claude Code in CI, structured data extraction) with exam-style questions, including "select TWO" items. Each question is tagged by domain so you can track your score, and we close with the ten reflexes that decide most questions.


AI-generated hosts. Independent study aid, not affiliated with Anthropic. The official Exam Guide is the source of truth.

CCAR-F #5 · Structured Output: Schemas Guarantee Structure, Not Truth05 oct. 202600:54:16

Domain 4 (20%). Explicit criteria over vague instructions, few-shot examples, structured output with tool_use and JSON schemas, nullable fields to stop fabrication, validation-retry loops, the Message Batches API (what it fits and what it does not), and multi-pass code review. The rule to remember: a schema guarantees structure, validation guarantees truth.


AI-generated hosts. Independent study aid, not affiliated with Anthropic. The official Exam Guide is the source of truth.

CCAR-F #3 · Tool Design and MCP: Fix the Description First05 oct. 202600:46:15

Domain 2 (18%). Why tool descriptions drive tool selection, structured errors with the MCP isError flag, why 4 to 5 tools per agent beats 18, how tool_choice works (auto, any, forced), MCP server scoping with .mcp.json and ~/.claude.json, MCP resources as content catalogs, and when to use Grep, Glob, Read, Write and Edit. We walk one tool call end to end and close with a design checklist.


AI-generated hosts. Independent study aid, not affiliated with Anthropic. The official Exam Guide is the source of truth.

CCAR-F #2 · Agentic Loops and Subagents: Hooks Beat Prompts01 oct. 202600:59:29

The heaviest domain on the exam (27%). The agentic loop and stop_reason, coordinator and subagent patterns, why subagents never inherit context, spawning with the Task tool, programmatic enforcement with hooks versus prompt guidance, task decomposition, and session resume and fork. Includes four mini exam questions with every distractor eliminated.


Study series for the Claude Certified Architect – Foundations exam. Audio generated with AI from personal study notes; not official Anthropic material. The official exam guide always wins on facts and weights: https://www.anthropic.com/learn/certification

CCAR-F #1 · Six Scenarios, 60 Questions: How the Exam Works01 oct. 202600:21:08

How the CCAR-F exam works and how to read its questions. We cover the format (60 items, 4 of 6 scenarios drawn at random, 120 minutes, scaled passing score of 720), the five domain weights, the six scenarios, and what is explicitly out of scope so you don't waste study time. Then a worked example: find the root cause, take the proportionate first step, and distrust over-engineered answers.


Study series for the Claude Certified Architect – Foundations exam. Audio generated with AI from personal study notes; not official Anthropic material. The official exam guide always wins on facts and weights: https://www.anthropic.com/learn/certification

CCDV-F #8 · Avoiding Critical Claude API Production Traps01 oct. 202600:40:40

The mistakes that never show up in the tutorial but always show up in production: error handling and retries (429, 500, 529), rate limits and retry-after, truncated responses, and context limits. A review to get you to the exam, and to production, without surprises.


Study series for the Claude Certified Developer – Foundations exam. Audio generated with AI from personal study notes; not official Anthropic material. The official exam guide always wins on facts and weights: https://www.anthropic.com/learn/certification

CCDV-F #7 · Hardening AI Agent Security Architectures01 oct. 202600:22:17

What to do when your agent reads content you can't trust. Prompt injection, secrets handling, PII, least privilege, and human approval for irreversible actions. A small slice of the exam (~8%), but a big deal in production.


Study series for the Claude Certified Developer – Foundations exam. Audio generated with AI from personal study notes; not official Anthropic material. The official exam guide always wins on facts and weights: https://www.anthropic.com/learn/certification

CCDV-F #6 · Adaptive Thinking and Model Selection: Opus, Sonnet or Haiku01 oct. 202600:05:26

How to pick the right model for the cost, latency and difficulty of the task, and how adaptive thinking works. Also what happened to budget_tokens on newer models, and why you shouldn't add date suffixes to model IDs.


Study series for the Claude Certified Developer – Foundations exam. Audio generated with AI from personal study notes; not official Anthropic material. The official exam guide always wins on facts and weights: https://www.anthropic.com/learn/certification

CCDV-F #5 · Do You Really Need an AI Agent?01 oct. 202600:24:41

Agents are fashionable, but a single call or a workflow is often enough. We discuss when to build each, and the 4 criteria that justify an agent: complexity, value, viability, and recoverable errors. If any one fails, start with something simpler.


Study series for the Claude Certified Developer – Foundations exam. Audio generated with AI from personal study notes; not official Anthropic material. The official exam guide always wins on facts and weights: https://www.anthropic.com/learn/certification

CCDV-F #4 · MCP: How to Give AI Hands01 oct. 202600:23:34

Model Context Protocol, demystified: servers, tools, resources and prompts, and how it all connects from the API. The part people forget most is that the connector needs two pieces, mcp_servers and an mcp_toolset that references it. Configure one without the other and Claude can't use the server.


Study series for the Claude Certified Developer – Foundations exam. Audio generated with AI from personal study notes; not official Anthropic material. The official exam guide always wins on facts and weights: https://www.anthropic.com/learn/certification

© My Podcast Data · Projet indépendant · Données issues d'Apple & Spotify