CourseAI / ML Engineer: The Complete Course · Module 15: Professional Practice · part 90 of 93
Part 90 · Module 15: Professional Practice

Lab: The Kestrel repository, ready for strangers

206 min read·
Module 15 · Lab · About 50 minutes · Needs: lessons 1 and 2 of this module, and the Module 14 lab (or nothing: Step 1 rebuilds Modules 1 to 14)

What you are building

Ready for strangers Measured, written, rerun from a clone Active flow 1 Processing step number Optional or fallback Your Drive the repository Modules 1 to 15 v2, measured value, latency artifacts/ not in Git What a stranger reads README.md numbers, sources kestrel-design.md a design answer A fresh clone git clone: only what Git holds same data f0423440d615 34 tests passed v2 rebuilt: 8c7d74ca8275 and its ONNX file 1 2 3
The lab measures v2 on your Drive, writes the README and the design answer a stranger reads, then reruns the README's commands in a fresh clone that holds only what Git holds (dashed: never committed); circles give the order.

Fourteen modules built one repository. This lab makes it something a stranger can judge in five minutes and rerun in thirty: a portfolio README.md whose every number has a source, a mock system-design answer in docs/interview/kestrel-design.md, and proof that the README's own commands work from a fresh clone. Lesson 2's warning applies: a repository that only reruns with its author's help is not evidence.

The headline numbers come from the real model v2, measured again here with Lesson 1's kestrel/value.py, not from the lessons' stand-ins. No cell calls a hosted model.

Open a new Colab notebook, call it module-15-lab.ipynb, and run each cell in order.

The rest of this course is yours to keep

This course is bought on its own, once, and stays readable afterwards, including the parts added to it later.

Buy this course · ₹399See what is inside

Or open everything with a plan