Casebook/Case 15
Serving & cost

Building a shared GenAI application stack

The layers around a language model: gateway, retrieval, orchestration, evaluation, safety, and observability.

Reported by the primary sourceFACT LAYER

What we can attribute directly

LinkedIn published its journey building a GenAI application technology stack.

The account describes reusable platform layers for application teams.

Read LinkedIn Engineering — GenAI stack Primary source · last checked 26 Aug 2026
01 · Problem & constraints

The operating envelope

Fast product iteration, multiple models, grounding, safety, evaluation, cost, and enterprise reliability.

Actors

Model teams, platform owners, operators, downstream product systems, and people affected by decisions.

Evidence

Versioned data, configs, traces, artifacts, deployments, and outcomes aligned on one timeline.

Failure cost

Every team reinvents prompts, retrieval, model access, evaluation, and production controls.

02 · Architecture reconstruction

Trace the system before naming the bug.

  1. 01

    Producers emit versioned data or model artifacts.

  2. 02

    A platform validates, computes, stores, schedules, or routes them.

  3. 03

    Training or inference consumes the exact declared version.

  4. 04

    Telemetry joins the decision to system, data, and model identity.

  5. 05

    Operators compare outcomes, stop conditions, and the last known-good path.

03 · Symptoms & investigation

Follow the evidence boundary by boundary.

Symptoms

Every team reinvents prompts, retrieval, model access, evaluation, and production controls.

Investigation

Trace prompt version, retrieval corpus, model route, policy checks, tool calls, citations, and feedback for one answer.

DIAGNOSTIC EXERCISE

An answer loses a citation after a model change. List every non-weight artifact needed to reproduce it.

Open investigation scaffold
  1. Write the earliest known-bad timestamp.
  2. Compare exact identities on either side of that boundary.
  3. Find the smallest affected slice and a known-good counterexample.
  4. Separate mitigation from root-cause confirmation.
04 · Root cause & fix

Repair the contract, not only the symptom.

ROOT CAUSE

Treating the provider call as the application leaves critical behavior unversioned and unobservable.

FIX

Standardize gateways, orchestration, retrieval contracts, evaluation, safety policy, and trace context.

Rollout

Start with a small set of high-value primitives; test escape hatches and version every behavior-changing layer.

05 · Rejected alternatives

Reason about the tempting shortcuts.

  • One giant framework that owns product logic.
  • Judging releases only by a general chatbot score.
06 · Monitoring after the fix

Make recurrence visible early.

01

End-to-end trace and cost

Define owner, slice, normal range, alert persistence, and the exact mitigation the alert should trigger.

02

Grounding/citation quality

Define owner, slice, normal range, alert persistence, and the exact mitigation the alert should trigger.

03

Safety and fallback outcomes

Define owner, slice, normal range, alert persistence, and the exact mitigation the alert should trigger.

REUSABLE PRODUCTION PATTERN

A GenAI platform manages evidence and behavior across many versioned layers.

Carry this pattern into assignments as a design constraint and into incident reviews as a hypothesis—not as proof about an unpublished system.