12 micro-topics in AI Systems, each with the evidence that proves you have it and the written reason behind every link. Before you start, it rests on 4 other domains. Downstream, it holds up 9 domains.
12 topics · depth 4–8 of 16 · 11 internal links · 9 in · 11 out
Load-bearing links first. Each one says what in this domain rests on what outside it, and why.
4 links
Configure a model client rests on Navigate the model landscape
You configure a specific model.
Tune generation parameters rests on Reason about nondeterministic output
Sampling parameters control the variation.
Maintain chat context rests on Explain the context window
History consumes the window.
Enforce spend caps in code rests on Compute token economics
A cap requires knowing what a call costs.
3 links
Apply a circuit breaker to model calls rests on Apply rate limiting and backoff
A breaker is the escalation of a retry policy.
Choose sync or async calls rests on Write async Python
The async path requires async Python.
Configure a model client rests on Call HTTP APIs asynchronously
Provider clients are HTTP clients with a wrapper.
1 link
Write system instructions rests on Write direct instructions
A system instruction is an instruction.
1 link
Get structured output rests on Define schemas with Pydantic
The schema is a model.
In prerequisite order. Each idea names the artifact that closes it, the market demand that put it on the map, and what it stands on.
Depth 4 · procedural · unassisted · M17 §17.1
Authenticates and instantiates a client
Navigate the model landscape · Model Foundations
You configure a specific model.
Call HTTP APIs asynchronously · Production Python
Provider clients are HTTP clients with a wrapper.
Depth 5 · conceptual · unassisted · M17 §17.1
Adjusts temperature and sampling deliberately
Parameters are set on a configured client.
Reason about nondeterministic output · Model Foundations
Sampling parameters control the variation.
Depth 5 · procedural · scaffolded · M17 §17.1
Sets model-side content controls
Safety settings are client configuration.
Depth 5 · conceptual · unassisted · M17 §17.2
Carries history across turns deliberately
Sessions carry state across calls.
Explain the context window · Model Foundations
History consumes the window.
Depth 5 · procedural · scaffolded · M17 §17.4
Sends non-text inputs to a model
Non-text input is sent through the same client.
Depth 5 · conceptual · scaffolded · M23 §23.4
Blocks calls that would exceed a budget
Compute token economics · Model Foundations
A cap requires knowing what a call costs.
The cap wraps the call.
Depth 6 · conceptual · unassisted · M17 §17.2
Matches call style to the workload
Both call styles use the same client.
Write async Python · Production Python
The async path requires async Python.
Depth 6 · representational · unassisted · M17 §17.2
Sets durable behavior separate from user input
System instructions are separated from turn history.
Write direct instructions · Prompt Engineering
A system instruction is an instruction.
Depth 7 · procedural · unassisted · M17 §17.2
Consumes a token stream from the provider
Streaming is the async call style.
Depth 7 · procedural · unassisted · M17 §17.3
Constrains responses with a response schema
Structured output is a request configuration.
Define schemas with Pydantic · Validation & Pydantic
The schema is a model.
Depth 8 · conceptual · scaffolded · M17 §17.3
Recovers when output does not match the schema
Failure handling assumes you asked for a schema.
Depth 8 · conceptual · scaffolded · M23 §23.4
Stops calling a failing upstream
Apply rate limiting and backoff · Production Python
A breaker is the escalation of a retry policy.
Handle schema validation failures
Both are failure paths around an unreliable call.
Everything downstream that names an idea here as a prerequisite, grouped by where it lives.
1 link
Bound loops with budgets rests on Enforce spend caps in code
Spend caps are the same mechanism at the loop level.
1 link
Carry structured state instead of transcripts rests on Maintain chat context
You replace history replay with state.
1 link
Use a model as a judge rests on Get structured output
A judge returns a structured score.
2 links
Trace LLM applications rests on Configure a model client
You trace model calls.
Track and control AI cost rests on Enforce spend caps in code
Caps and attribution are two halves of budget control.
1 link
Write direct instructions rests on Configure a model client
You need something to prompt.
1 link
Inject retrieved context rests on Maintain chat context
Retrieved text is placed into the request.
1 link
Implement content safety filters rests on Configure safety settings
Provider settings are the first filter; yours is the second.
1 link
Declare tools for a model rests on Get structured output
Tool arguments arrive as structured output.
2 links
Manage conversation history in the UI rests on Maintain chat context
The UI mirrors how context is carried on the server.
Handle multimodal file input rests on Process images and documents
Upload UI exists to feed multimodal input.
Retool. Retrain. Relaunch.
375 ideas. 17 weeks. No tuition, ever.
Vets Who Code is a veteran-run 501(c)(3). The accelerator is free, remote, and we don’t take a share of your first paycheck.
Free · Remote · 17 weeks · EIN 86-2122804