ML Engineering on AWS
Outcome: own ML systems end-to-end
This path is not about learning machine learning theory — it is about owning an ML system end to end on real infrastructure: training on SageMaker, serving a live endpoint, running an A/B test between model variants, watching for drift, and wiring retraining into a pipeline that redeploys itself. If you can already train a model locally but have never shipped one behind an endpoint, monitored it, or paid an AWS bill for it, this is the gap this path closes.
Several of these projects spin up real AWS resources that cost money while running. Every one of them ships with a teardown script — run it the moment you are done testing.
Train an XGBoost model locally, then submit the exact same script to SageMaker. The point is not a new algorithm — it is seeing that the same training code runs in both places, so “local” and “cloud” stop being two different skill sets and become one workflow with a different target.
Proves you can move a training job from a laptop to managed cloud infrastructure without rewriting it — the baseline expectation for any ML Engineer role.
What's next
Everything in this path trains and serves classical ML models — none of it touches large language models directly. Path 1 picks up exactly there, building real LLM-backed tools from scratch. If you would rather apply AI inside an existing software engineering role instead, Path 2 is the other independent option. Neither requires finishing this path first. AI Engineering Fundamentals →