A Berlin Release Shift: Learning Rust Before Friday

1. Pick a task small enough to finish

It is Tuesday morning in Berlin, and Friday’s release needs a tiny command-line tool that reads a CSV file and reports invalid rows. You know Python, but the team’s utility scripts are written in Rust. That concrete task gives your learning a boundary: understand enough Rust to build and test this one tool, not enough to memorize a language reference.

Write down the input, expected output, and two failure cases. For example, accept a file path, print a count of invalid rows, and report a useful error when the file is missing. Keep this checklist beside your editor. It will help you decide what to study and when to stop.

2. Ask AI for a short learning map

Give an AI assistant your task, your existing programming experience, and a time limit. Ask it to identify the language concepts you will need, order them by dependency, and suggest small exercises before showing a complete solution. For this Rust tool, that list might include variables and types, ownership, reading files, parsing records, error handling, and tests.

Ask for unfamiliar terms to be explained with small examples, and request links or search terms for the official documentation. Treat the model as a tutor, not an authority: verify syntax and behavior against the compiler and language docs. If you use RelayVano for model access, choose a model that fits your session, such as Claude Sonnet 5.5, and keep the same task context in your prompts.

3. Build a baseline before reading a full solution

Open the project and get the smallest program compiling. Then attempt one feature at a time, even if your first version is rough. When you get stuck, paste the smallest relevant snippet and the exact compiler error. Ask the assistant to explain what the error means, name the concept behind it, and offer a hint before proposing a fix.

That sequence matters. A complete answer may make the immediate problem disappear while leaving you unable to recognize it next time. A hint gives you a chance to form a hypothesis, change the code, and see what the compiler says.

4. Turn every explanation into a prediction

After the assistant explains a concept, close the answer and predict what a small code change will do. In Rust, you might ask yourself whether moving a string into a function leaves it available afterward, or what happens when a file cannot be opened. Then run the code or a focused test to check your prediction.

When your prediction is wrong, ask for a comparison between your expectation and the actual behavior. Request one new example that changes only a single detail. This keeps the lesson focused and helps you build a reliable mental model instead of collecting disconnected explanations.

5. Keep a prompt and mistake log

Maintain a short text file with three sections: concepts you can explain, errors you have fixed, and questions to revisit. For each mistake, record the compiler message, its cause in your own words, and the change that resolved it. Ask the assistant to quiz you from this log without revealing the answers first.

Use the log to shape the next session. If ownership keeps causing confusion, spend fifteen minutes on that topic before adding another feature. If a concept already feels clear, test it by explaining it without notes or applying it in a different part of the tool.

6. Finish with a code review and a clean rebuild

Once the tool meets your checklist, ask AI to review it for missing error cases, confusing code, and tests you have not written. Request findings with file and function names, not an unexplained replacement implementation. Review each suggestion yourself and run the project’s formatter, tests, and build commands from a clean state.

Before Friday, rebuild the tool from a fresh directory or explain each key section aloud while reading the code. Mark any line you still cannot explain, then investigate it in the documentation or with a targeted question. You have learned the language usefully when you can make a change, predict its effects, and debug the result.

7. Keep the AI session available in your workflow

For a repeatable setup across coding tools, RelayVano provides pay-as-you-go access to AI models and works with Claude Code, Codex, Gemini CLI, and OpenAI-compatible SDKs at https://api.relayvano.com. Sign up for RelayVano when you are ready to connect a model to the workflow you already use. Keep your prompts scoped, verify generated code locally, and let the compiler and tests decide whether the lesson holds up.