SVM Engineering

Lessons

1. OrientationWho we are, how the six weeks run, your own agent and your portfolio.
2. Your development environmentUse the provided workflow file to configure VS Code and Rust Analyzer for exploring the Agave repository—without building the validator. Learn to navigate a large codebase and get started with OpenCode 2.
3. AI hygieneWhat leaves your machine when you work with a coding agent, and what should not.

You hand in

• Toy program, written by agents, explained by you

Lessons

1. From idea to planFive toy programs presented, then the plan: the difference between a vibe coder and an agentic engineer.
2. Your capstone: choose and specifyChoose a capstone close to the Agave validator, and specify it before any code is written.
3. Benchmarking and the DashMap questionMeasure before you believe: synchronisation primitives under load, and why a fast map can still be the wrong choice.

You hand in

• Plan first, then rebuild your toy
• Capstone specification
• Benchmark synchronisation primitives and explain the winner

Lessons

1. Make your agent explainHow your designs went, and how to make an agent teach you the why behind each choice.
2. Capstone plans, presentedPresent your capstone plan, answer questions, and sharpen it before you build.
3. The schedulerWhat the transaction scheduler is, what it does, how it works, and what it makes possible: block ordering, BAM, schedulers outside the validator.

You hand in

• Refine your capstone plan with your agent
• Capstone repository
• Toy: shaq and rts-alloc the way the scheduler bindings use them

Lessons

1. Alpenglow: Solana's new consensusWhat Alpenglow is, what it replaces, how Votor and Rotor work, and where they live in the Agave codebase.
2. XDP: the kernel bypassWhat XDP is, how it made the Agave client faster, the code behind it, and what you would change.
3. eBPF: inside a running systemLook inside a running Linux program with eBPF: what it does, how a profiler works, and what you would build.

You hand in

• XDP: a toy program in a Linux VM
• eBPF: a toy project with aya

Lessons

1. Your capstone: the architecture and the technologiesYou are building it. Now you explain it: the parts, the arrows, and the technologies you chose.
2. Your capstone: the decisions and the trade-offsThe decisions you made, the alternatives you rejected, and what each one cost you.

You hand in

• Capstone: the architecture and the technologies
• Capstone: the decisions and the trade-offs

Lessons

1. Explain your capstoneThe last explanation before the demo: the architecture, the technologies, the decisions and the trade-offs, as one story.
2. Demo dayThe last class of the cohort: you run it live, you defend it, and you write down what you learned.

You hand in

• Demo: the script
• Final: the capstone and what you learned