Leon Jose

Leon JoseNov 1, 2026

Learning AI feels easy until you realise how much you don't know. Day 1: Write your first ChatGPT prompt. Day 30: RAG, MCP, AI agents, evals, context engineering... and 10 new models launched this week. Nobody tells you how quickly the AI rabbit hole gets deeper. You start with a simple prompt. Then you discover: → Prompt engineering. → RAG and embeddings. → MCP and tool calling. → AI agents and workflows. → Context engineering and evaluations. → Fine-tuning and model architectures. And just when you think you've caught up, something new launches. But here's the thing: You don't need to learn everything at once. Pick one concept. Build something with it. Understand how it works. Then move to the next. The goal isn't to know every AI tool. It's to know how to use AI to build things that matter. Anyone else experiencing this? 😭 --- ♻️ Repost if your AI learning journey looks like this. ➕ Follow Leon Jose for more AI educational content

Leon JoseOct 31, 2026

A Google engineer explains Agentic harnesses in 9 min, better than any other tutorial I've seen. Annie Wang walks through how to setup an Agentic harness. AI model alone isn't an agent. It needs software around it to provide context, connect tools, enforce permissions, and manage execution. The harness is the operating system around the model. Save this before your next build: 𝟭/ 𝗖𝗼𝗻𝘁𝗲𝘅𝘁 𝗮𝗻𝗱 𝗺𝗲𝗺𝗼𝗿𝘆. ↳ Sessions store the conversation, tool calls, and results. ↳ State tracks important information the application needs during a task. ↳ Long-term memory carries useful preferences into future conversations. 𝟮/ 𝗧𝗼𝗼𝗹𝘀 𝗮𝗻𝗱 𝗲𝘅𝗲𝗰𝘂𝘁𝗶𝗼𝗻 𝗳𝗹𝗼𝘄. ↳ Tools let agents perform actions, from editing files to calling external APIs. ↳ MCP servers provide a standard way to connect external tools. ↳ Workflows control execution, parallel tasks and retries. 𝟯/ 𝗥𝘂𝗻𝘁𝗶𝗺𝗲, 𝘀𝗲𝗰𝘂𝗿𝗶𝘁𝘆 𝗮𝗻𝗱 𝗽𝗿𝗶𝘃𝗮𝗰𝘆. ↳ Sandboxes isolate the agent and restrict what it can access. ↳ Guard against prompt injection with restricted permissions, protected credentials, and human approval for sensitive actions. ↳ Control where data goes, what gets stored, and how long it's retained. 𝟰/ 𝗔𝗴𝗲𝗻𝘁 𝗲𝘃𝗮𝗹𝘂𝗮𝘁𝗶𝗼𝗻. ↳ Code checks whether the output meets requirements. ↳ LLM-as-a-judge evaluates qualities like clarity and completeness. ↳ Human evaluation catches issues automated checks might miss. Here's the example that ties everything together: You ask an agent to build a CSV export feature. It reads the relevant files, edits the code, runs tests, finds a missing customer record, and tries again. The harness manages that entire process. That's the part most people overlook. A better model isn't the only answer. You need the right context, tools, execution flow, security controls, and evaluation around it. ♻️ Repost. Your network will thank you. ➕ Follow Leon Jose for more AI educational content

Leon JoseOct 30, 2026

You don't need a $500 Claude course. Anthropic launched Claude Academy. It's FREE. Inside: 22 courses, 119 tutorials, 148 use cases, 66 webinars. START HERE Claude 101 → https://lnkd.in/engztGZA AI Fluency → https://lnkd.in/es95MxXY AI Capabilities and Limitations → https://lnkd.in/ecb5ZV4u LEARN CLAUDE TOOLS Introduction to Cowork → https://lnkd.in/e_Tqs7bG Claude Code 101 → https://lnkd.in/exBE7WM9 Claude Code in Action → https://lnkd.in/eD5kj28a Introduction to Agent Skills → https://lnkd.in/e8dfeGzn Introduction to Subagents → https://lnkd.in/ez5itUPy LEARN AI FOR YOUR ROLE AI Fluency for Businesses → https://lnkd.in/eZHqKTzR AI Fluency for Builders → https://lnkd.in/eyJnEK28 AI Fluency for Students → https://lnkd.in/ec_GRWyy AI Fluency for Educators → https://lnkd.in/egG-GyGy AI Fluency for Nonprofits → https://lnkd.in/epM6NwmE BUILD WITH THE CLAUDE API Claude Platform 101 → https://lnkd.in/eDpzFxcm Introduction to MCP → https://lnkd.in/euYZzU9Z MCP: Advanced Topics → https://lnkd.in/eAwUQjbc Building with the Claude API → https://lnkd.in/ex7SNKuW Claude with Amazon Bedrock → https://lnkd.in/ewbRywdM Claude with Google Vertex AI → https://lnkd.in/eFEr4CKp THE OTHER 333 119 tutorials → https://lnkd.in/e2r7g5Di 148 use cases → https://lnkd.in/exBB69ZP 66 recorded webinars → https://lnkd.in/eCh-2U3D Free to access (No payment required) Image credits: Charlie Hills --- P.S: Join my free AI newsletter Weekly free resources + practical AI tutorials. Subscribe here 👉 aiforcareer.co --- ♻️ Repost to help someone in your network