Many of us already use Claude Code, Codex, or similar coding agents for daily work: asking questions, writing docs, debugging issues, making small code changes, or reviewing code. This training helps you get much more productivity out of the agents you already use.
We will start with writing, research, documentation, and websites, so the first half is useful even if you do not write code every day. You will learn how to use agents to draft, critique, revise, match your style, and publish work to the web.
Then we will move into deeper engineering workflows: repo exploration, code review, performance testing, deployment, and longer-running agent loops. The capstone is intentionally unfamiliar: building a Rust module for Valkey. You do not need Rust or Valkey experience going in. The point is to learn how to use agents to move through unfamiliar systems with more confidence.
Course Outline
1. Build a Better Working Relationship with Your LLM
We will start with the journey of a prompt: how your request is handled, what happens from turn to turn, and why context matters. You will learn how to make better use of memory, project instructions, and lightweight conventions so your agents understand your goals, preferences, and working style.
2. Write, Publish, and Style Work with Agents
You will use Claude Code to research, draft, revise, and publish a short piece of writing. We will use multiple agents for research, synthesis, devil’s advocate review, editing, and style iteration. Then we will turn the work into a simple website, even if you have never built a site or written JavaScript.
3. Build Technical Fearlessness
We will build a simple Valkey module in Rust. Ideally, this is unfamiliar territory. That is the point. You will use agents to explore the codebase, understand the APIs, implement the module, write documentation, and explain what you built. This is where you will get your hands dirty and learn how to produce and understand code. You will use the process to learn about Valkey modules and walk away understanding more about the modules by building them.
4. Run Performance Tests Without Living in Unix Commands
You will learn how to use agents to run performance sweeps, vary parameters, collect results, identify bottlenecks, document findings, and propose optimizations. You will learn how to let the agent handle the mechanical work while you focus on judgment.
5. Use Agent Loops and Agent-to-Agent Workflows
Finally, we will move beyond one prompt at a time. You will learn how to set goals, define success criteria, run review loops, and use agents that prompt, critique, and check each other’s work until the code, docs, tests, and website are complete.
Hands-On Work
- Create a writing workflow that matches your voice
- Build and revise a small website (product page) with an agent
- Set up useful memory for project and personal context
- Use Claude Code to explore and modify a real codebase
- Use Codex as an independent reviewer
- Run a benchmark or performance sweep
- Build, test, review, document, and deploy a Rust Valkey module
Key Takeaways
1 Use Claude Code and Codex for writing, documentation, websites, and engineering work
2 Build reusable workflows for drafting, critique, revision, testing, and deployment
3 Make better use of memory so agents learn your project, preferences, and style
4 Use agents to operate the shell without writing every Unix command yourself
5 Run repo exploration, implementation, code review, and performance sweeps with agents
6 Use multiple agents together, including devil’s advocate review and agent-to-agent workflows
7 Build confidence using agents on unfamiliar systems
Speaker
Khawaja Shams
Co-Founder & CEO @Momento, previously @NASA and @Amazon
Khawaja is a passionate QCon advocate. He earned the NASA Early Career Medal for his contributions to the Mars Rovers, from the cameras to the image processing pipeline. At Amazon Web Services, he ran DynamoDB and was subsequently the VP of Engineering for AWS Elemental. Now, as CEO and Co-founder of Momento, he's leading a team that's building serverless platforms to empower developers with caching and pub-sub.
