---
title: "Preferences > Benchmarks: Model Routing for How Teams Actually Build"
category: "talks"
date: "2026-07-01"
time: "12:05pm-12:25pm"
track: "AI Architects: AI Factories"
room: "Leadership 2"
speakers: ["Archana Kamath", "Tyler Gillam"]
sourceLabels: ["Official conference schedule", "Public YouTube metadata"]
scheduleTrack: "AI Architects: AI Factories"
scheduleRoom: "Leadership 2"
scheduleLabels: ["AI Architects: AI Factories", "Leadership 2", "session", "confirmed"]
---
# Preferences > Benchmarks: Model Routing for How Teams Actually Build

## Conference Context
- Date/time: 2026-07-01 · 12:05pm-12:25pm
- Track/room: AI Architects: AI Factories · Leadership 2
- Speaker(s): Archana Kamath, Tyler Gillam
- Session type/status: session · confirmed

- Track: AI Architects: AI Factories
- Room: Leadership 2
- Session type: session
- Status: confirmed

## Session Description
There is no best model. There's only the right model for a given task, and the right model depends on your team's preferences, not a benchmark score. This talk makes the case for preference-aligned routing: choosing models by the constraints that actually matter — cost, latency, task type, model preference — instead of a single leaderboard number. We'll demo a sub-200ms routing decision running on a purpose-built 30B MoE model with no application code changes, walk through real coding workflows routing most traffic to open models without losing accuracy, and show where this goes next: evals, caching, and personalization.

## Media Evidence
No related AI Engineer channel video found yet.

## Evidence Graph
This evidence graph is generated from currently linked source material: official schedule text, related video pages, cached transcripts, visible slide text, dense/reconstructed slide pages, and AI slide-classification audits.

### Media Signals
No linked video, transcript, or slide source has been attached yet.

### Agent Reading Notes
Use these signals to refine the synopsis, topic links, people/company context, and method notes. If a source is a related external video rather than an exact official recording, keep it framed as supporting evidence.

## Transcript Status
No official session recording transcript was found by exact title match on the AI Engineer YouTube channel during this run.

## People
- [[archana-kamath]]
- [[tyler-gillam]]

## Notes
- Pending transcript synthesis when an official recording or confirmed matching video is available.

## Synthesis
### Synthesized Breakdown
# Preferences > Benchmarks: Model Routing for How Teams Actually Build ## Conference Context - Date/time: 2026-07-01 · 12:05pm-12:25pm - Track/room: AI Architects: AI Factories · Leadership 2 - Speaker(s): Archana Kamath, Tyler Gillam - Session type/status: session · confirmed - Track: AI Architects: AI Factories - Room: Leadership 2 - Session type: session - Status: confirmed ## Session Description There is no best model. There's only the right model for a given task, and the right model depends on your team's preferences, not a benchmark score. This talk makes the case for preference-aligned routing: choosing models by the constraints that actually matter — cost, latency, task type, model preference — instead of a single leaderboard number. We'll demo a sub-200ms routing decision running on a purpose-built 30B MoE model with no application code changes, walk through real coding workflows routing most traffic to open models without losing accuracy, and show where this goes next: evals, caching, and personalization.

### Speaker And Company Context
- [[archana-kamath|Archana Kamath]] — VP of Engineering at [[digital-ocean|Digital Ocean]].
- [[tyler-gillam|Tyler Gillam]] — Senior Software Engineer II - Agentic AI at [[digital-ocean|Digital Ocean]].

### Topics Covered
- [[coding-agents]]

### Derived Links And Source Material

### Novel Concepts / Clever Methods
- No highlighted novel concept has been detected yet.

### Evidence Boundary
This synthesis is based on the official schedule and linked source pages. It should be revisited when exact session recordings or transcript-backed secondary sources are available.