SignalNest's Model Menu

The situation

SignalNest's lead engineer demoed a prototype to the board: 'an AI that summarises any document'. It's built on AWS. The board is excited and Leo has already used the word 'transformative' twice. Amara wants a one-page comparison that says what it actually is, what it's good and bad at, and which AWS service it should be built on — in plain business terms, without a line of code and without overselling it.

What you'll learn

• AIF-C01 AI/ML fundamentals: predictive vs generative AI, the ML lifecycle, foundation models and generative AI (benefits and limits), and choosing among AWS AI options — pre-built AI services, managed foundation models (Bedrock), and custom model training/hosting (SageMaker) — for a specific need.
• The AI, ML and deep learning hierarchy; supervised, unsupervised and reinforcement learning; data types and the four kinds of inference; the ML lifecycle and MLOps; performance and business metrics (accuracy, precision, recall, F1, AUC); and which SageMaker AI capability serves each lifecycle stage, plus PartyRock and token-based versus provisioned pricing.

Who you work with

• 

  Amara Okafor

  ML and Data Engineer

• 

  Priya Naidoo

  Cloud Architect

Scenes

1. 1.'An AI that summarises anything'
2. 2.Predictive or generative
3. 3.Which AWS service
4. 4.Which SageMaker AI feature?
5. 5.Build or buy
6. 6.AI fundamentals check

What you leave with

The fictional board brief recommends evaluating a managed foundation model rather than building from scratch under the supplied constraints. It records integration and evaluation work, acknowledges unsupported-output risk and rejects training on every company document without an authorised dataset or measured gap.