← All drillsCall an LLM Like an Engineer
What you'll be able to do- ✓Call a hosted language model from Python and read every field of the reply, including why it stopped
- ✓Count the tokens in any text and predict what a prompt will cost, instead of estimating from words
- ✓Budget a context window across system prompt, history and hidden thinking, and find the limit you actually hit first
- ✓Explain why a model has no memory between calls, and compare full-history, window, summary and pinned strategies on cost and on what they forget
- ✓Measure what temperature, top-p and a seed really guarantee by repeating a prompt and counting the distinct answers
- ✓Stream a reply and time the first event, the first visible word and the end, and say when streaming helps
- ✓Tell a rate limit from an oversized request, a bad key and a retired model, and retry only what is worth retrying
- ✓Build a chat app that streams, meters its own cost, recovers from cut-off answers and keeps its history affordable, and compare two models on cost per correct answer
- ✓{'Read a call log (JSONL) like an engineer': 'sort its lines by shape, total tokens and cost, find silent truncation, the slow tail and the failures, and avoid double-counting the lines of an agent transcript'}
Make a reliable, costed first call to a language model from your own code, the base every chatbot, copilot and agent feature is built on.
⌁ Every product that calls a language model API, from support chatbots to coding assistants and in-app copilots.
Start this internshipCreate an account to unlock the 10 sections, the workbench, and askThili.
BeginSections
1. Call an LLM Like an Engineer
🔒 locked2. Your first call
🔒 locked4. The context window is a budget
🔒 locked5. The model has no memory
🔒 locked7. Streaming and latency
🔒 locked8. When calls go wrong
🔒 locked9. A chat that meters itself, and which model to use
🔒 locked10. Reading the log like an engineer
🔒 lockedDig deeper
🔗Groq documentation (the API used in this drill)
docs🔗thilillm, the reference implementation you check your code against
codePart of these learning paths
I'm a student and I want to get my first AI/ML job
View path →