Operators, product owners, and people affected by the foundation models decision.
Transformers · weeks 20
Foundation-model lifecycle
A private support assistant must answer from evidence and resist manipulation.
Opening production brief
A measurable problem, before a model.
build vs buy, RAG vs tuning, provenance, decontamination with version, owner, event time, and availability contracts.
A private support assistant must answer from evidence and resist manipulation. Every solution must state latency, cost, capacity, and fallback limits.
Offline evidence plus a deployable resolution: A grounded system design spanning adaptation, RAG, evaluation, and safety.
Frame and baseline
Turn the production problem into explicit data, metric, and baseline contracts.
- 01↗
Prompt, retrieve, adapt, train, or buy?
Match the intervention to the failure: prompts shape instructions, retrieval updates knowledge, tuning shapes behavior, and training or buying changes base capability.
65–90 min · checkpoint · deterministic lab - 02↗
Corpus provenance, mixtures, and decontamination
Corpus quality is governed by provenance, realized sampling, and auditable exclusions—not merely by token count.
65–90 min · checkpoint · deterministic lab
Build and stress
Formalize the machinery, test counterexamples, and expose system limits.
- 03↗
Pretraining objectives, scaling, and checkpoints
Pretraining progress is tokens transformed into validated capability, backed by slice-aware loss and checkpoints that preserve the exact continuation state.
65–90 min · checkpoint · deterministic lab - 04↗
SFT, preference optimization, safety, and tools
Post-training aligns behavior only to the examples and constraints you encode; tool permissions and safety must still be enforced and evaluated outside the loss.
65–90 min · checkpoint · deterministic lab
Resolve and operate
Join model behavior to architecture, observability, rollout, and rollback.
- 05↗
Evals, model cards, feedback, and reproducibility
Reliable releases bind artifacts, predeclared eval gates, honest limitations, and bias-aware feedback into one reproducible decision record.
65–90 min · checkpoint · deterministic lab - HW↗
Domain adaptation release package
Compare prompt, retrieval, and optional PEFT approaches under one provenance and evaluation contract.
Rubric · staged hints · reference resolution
Prerequisite guidance
1 concepts make this faster.
Prerequisites are guidance, never hard gates.