Agentic Coding
Essentials
Master AI coding agents, build AI-powered applications, and ship production-ready agent systems - with dual Python and JavaScript tracks.
Whether you're ready to get up to speed with AI development, leading a team that needs to ship faster, or a vibe coder ready for the next level - this course meets you where you are and takes you further.
Hi, I'm JV.
I'm passionate about helping people make great software.
In my 25 years of coding and 12 years teaching development, I've never seen anything impact my profession as much as AI.
It has changed how I build, what I build, as well as how I teach and learn.
My main focus for the last three years has been keeping up with AI and how it is changing my profession.
I would love to help you navigate this shift so you can stay ahead of the curve.
Go Deeper
Maybe you're an experienced developer ready to get current with AI tooling fast. Maybe you're leading a team and need them productive with AI yesterday. Or maybe you've been vibe coding and want to take your practice to the next level.
This course starts from the foundations and takes you all the way to building production agent systems. You'll learn the strengths and limitations of tools like Claude Code, Codex, and Antigravity CLI, how to structure projects so agents do their best work, and how to build your own AI-powered applications.
Here's the thing: best practices in software engineering are changing fast right now. Nobody has all the answers. The patterns that worked six months ago are already being replaced. Everyone is experimenting, and the craft is evolving in real time.
In this course, I'll share what some of these emerging practices look like right now - and more importantly, help you build the habits and mental models to discover and shape new practices yourself as this rapid evolution continues.
You'll transform both sides of your practice:
- How you work - using agentic coding tools with depth and intention
- What you build - creating intelligent applications with tools, structured outputs, and agent frameworks
Here's what you'll be able to do
After This Course You'll:
- 1.Use agentic coding tools like a pro - Claude Code, Codex, Antigravity CLI, with effective context engineering and project management workflows
- 2.Build AI-powered applications using LiteLLM (Python) or Vercel AI SDK (JavaScript) with multiple providers and local models
- 3.Extend your tools with MCP servers and skills - use existing ones and build your own custom servers
- 4.Ship production-safe AI systems with guardrails, observability (Langfuse), and red-team testing (Promptfoo)
- 5.Build tool-calling agents with structured outputs, grounding, and multi-agent coordination patterns
- 6.Ship a capstone agent system using Google ADK (Python) or Vercel AI SDK (JavaScript) that brings everything together
Python & JavaScript Tracks
Every hands-on exercise comes in both Python and JavaScript. Choose your track and learn the same concepts with industry-standard tools for your language.
Course Roadmap
You'll build one project across the entire course - a chatbot that evolves from a simple LLM wrapper into a full agent system with tools, MCP integration, guardrails, and observability.
Each module mixes lectures, live demos (real workflows, real mistakes), hands-on tutorials, open-ended challenges, and research projects - in your choice of Python or JavaScript.
Module 1: Agentic Coding Foundations
- The AI coding spectrum: autocomplete → inline assist → chat → agentic
- CLI tools deep dive: Claude Code, Codex, Antigravity CLI, GitHub Copilot CLI, OpenCode
- Context engineering: CLAUDE.md, AGENTS.md, rules files, memory and handoff patterns
- Agentic workflows: plan-execute patterns, git workflows with agents, backlog management with gh CLI
Module 2: Building with LLMs
- LLM fundamentals: tokens, context windows, temperature, provider pricing
- Python: LiteLLM unified API | JavaScript: Vercel AI SDK
- Build a chatbot (Streamlit or Next.js) - your foundation project for the course
- Local models with LM Studio and personal project kickoff
Module 3: Tools, Agents & Grounding
- Tool calling and function calling with LiteLLM or Vercel AI SDK
- Structured outputs with Pydantic (Python) or Zod (JavaScript)
- Agent patterns: augmented LLM, prompt chaining, routing, orchestrator-worker, evaluator-optimizer
- Multi-agent coordination and grounding with Gemini Grounding API
Module 4: Skills, MCPs & Extensibility
- MCP fundamentals: the Linux Foundation standard for giving agents tools, resources, and prompts
- Using MCP servers: filesystem, GitHub, Playwright, databases
- Agent skills: the SKILL.md open standard, installing and authoring skills
- Build a custom MCP server with the Python or TypeScript SDK
Module 5: Guardrails, Observability & Production
- The risks: prompt injection, PII leakage, hallucination, OWASP Top 10 for LLMs
- Input & output validation: Guardrails AI + Pydantic (Python) or Zod + Vercel AI SDK middleware (JS)
- Observability with Langfuse: tracing, cost monitoring, measuring guardrail effectiveness
- Red teaming with Promptfoo: adversarial testing and CI/CD safety checks
Module 6: Agent Frameworks & Capstone
- Python: Google ADK - agent types, tools, sub-agents | JS: Vercel AI SDK agents
- Build a real agent with MCP servers, skills, tools, guardrails, and observability
- Capstone project: ship an agent system that demonstrates everything you've learned
- Demo, code review, retrospective, and next steps
Choose your ideal learning tier
Master agentic AI development
Plus
Learn with a live community
$397
- Everything in Essentials
- Weekly live group calls with Q&A (Thursdays 12:30pm NZ time, starting April 2nd)
- Custom lessons based on group interest
- Discord community access with help desk support
Enrolments close Friday 3rd April
All prices are in USD - contact us if you want to pay by bank transfer
Meet the teaching team
I've been working with this handpicked team of Dev Academy alumni on commercial projects for over a year.
Each excels at learning, teaching, coding, and AI development – and I'm excited to now be teaching alongside them to help you bridge the gap between AI no-code tools and custom development.
Joshua Vial
JV studied computer science in the late 90s, has been programming professionally ever since. He co-founded Dev Academy in 2013, and has been consulting and teaching independently since 2023.
JV: you've already heard enough about me, I'm thrilled to introduce you to the team.
Jen Hurley
Jen both studied and taught at Dev Academy in 2023, and loves creating inclusive learning environments that demystify software development.
They meet learners where they are, encouraging exploration and making technology accessible through a supportive approach that transforms "I can't code" into "I just haven't coded yet."
JV: I've had to bribe the fundsorter.comteam to make some of Jen's time available for teaching. They are big fans.
Joel Anderson
Joel graduated Dev Academy in 2025, bringing years of experience in product management and scaling small business operations before retraining as a full-stack developer.
Joel has a background in industrial design and digital manufacturing.
JV: He's also great at coaching self-confessed hoarders (aka me) to get rid of junk in their shed. My wife marvels at his persuasive abilities.
Got Questions?
How much time do I need?
3 - 5 hours per week. On the Core plan you can get by with three hours per week but you will likely leave a few things to come back to. You have lifetime access to the course materials.
For Premium plans the time requirements can be customised to your availability.
Who is this course for?
This course is for three kinds of people: developers who want to get up to speed with AI coding tools, engineering leaders who want their teams shipping faster with AI, and vibe coderswho've been building with AI and want to understand what's happening under the hood.
You don't need to be a professional developer. If you've been writing code - whether for 20 years or 2 months with Cursor - you'll get value from this course. We start from foundations in Module 1 and build up from there.
Can I enroll my team?
Yes. If you're enrolling multiple people, get in touchand I'll set up team pricing and a shared coaching structure that works for your group.
Python or JavaScript - do I need to know both?
No. Pick the track that matches your experience. Every exercise comes in both languages, teaching the same concepts with the best tools for each ecosystem.
Do I need to pay for any AI tools?
Not unless you want to. Claude Code, Antigravity CLI, and LM Studio all have free tiers, and you can complete the course without spending money on tools.
Paid tiers remove rate limits and unlock more capable models. One goal of the course is to help you understand the landscape so you get the most value for your money if you decide to spend.
Is there a guarantee?
Yes, I offer a 14-day satisfaction guarantee. If you're not completely satisfied with the course content after two weeks, I'll provide a full refund.
