Nobody owns a cost they can't see.

AI Infrastructure & Unit Economics. I'm Tim Urista, a Senior Cloud Engineer at Apple — I make compute fleets legible and their costs attributable, from $10–12M/quarter of cloud spend to the per-signal economics of the AI systems I run myself.

Ask AI About My Experience

Search my work, projects, and writing for specific insights

Get the AI Readiness Kit

90-day AI implementation plan + readiness checklist + failure mode worksheet

No spam. Unsubscribe anytime. Used to send your kit and occasional AI insights.

Three Scales, One Discipline

🧾

Cost Attribution at Scale

Fleet-wide financial attribution across ~700K servers and $10–12M/quarter of cloud spend. Cost data as an engineering artifact, not an accounting afterthought.

📉

AI Unit Economics

Fully-loaded cost per inference, per request, per signal. I run production AI whose invoice I personally pay — ~$1.50–2.00 per signal on Trendvesting.

🔍

Legible Infrastructure

Internal platforms like cloudOS (~400 weekly users) that make livability, deployment state, and spend visible to Finance, SRE, and executives alike.

$0–
Quarterly Cloud Spend
Attributed at Roku
0K
Servers
Fleet-scale attribution at Apple
$0.
Per AI Signal
Fully-loaded, Trendvesting
~0
Weekly Users
cloudOS internal platform

See the discipline, live

Not a demo chatbot — a working cost-attribution tool. Describe your workload; get a number you could take to finance.

🧾

AI Cost Teardown

Describe an AI or cloud workload in plain English. You'll get a fully-loaded monthly estimate, a per-unit cost, and the highest-leverage ways to cut it. Live, powered by Claude.

Selected Work

View All Cases

Fleet-Scale Cost Attribution — Apple

Financial attribution pipeline (CIBO/FinTrack) correcting misattribution across ~700K servers, where quiet errors distort budgets by millions. Now piloting LLM inference to pre-label project cost hierarchies.

700K serversFleet-scaleLLM pre-labeling
Cost attributionPrivate cloudLLM

Cost-Attribution Pipeline — Roku

Built the pipeline tracking $10–12M/quarter of AWS spend and co-led the ECS→Kubernetes migration. Cost data treated as an engineering artifact, not an accounting afterthought.

$10–12M/qtrECS→K8s
AWSKubernetesCost attribution

Trendvesting — Unit Economics in Production

AI signal-intelligence platform live since Feb 2024. A real AI system whose invoice I personally pay — ~$1.50–2.00 fully loaded per signal, ~$400–500/month steady state.

Since 2024$1.50–2.00/signal
GoNext.jsFastAPI

TendForm — Infrastructure Discipline, Solo

My solo SaaS form builder with multi-form packets, a markdown DSL, a production MCP integration, and a HIPAA-compliant tier on a CloudNativePG enclave with a full BAA chain. Run like the fleet work, just smaller.

HIPAA tierMCP serverReal customers
React/ViteFastAPIKubernetes

Latest Writing

All Articles on Medium

The FinOps Tax on AI Engineer Salary Premiums

Why the cost of AI talent is itself a line item most orgs never attribute — and what that hidden tax does to unit economics.

Read on Medium → DevSecOps & AI · 7 min

Unexpected Costs of Fragmented AI Governance

AI governance sprawl is a hidden cost center: every disconnected policy, tool, and review adds spend nobody attributes back to it.

Read on Medium → DevSecOps & AI · 11 min

Beyond Tokens: How BLT Solves the $10B Language Model Crisis

Tokenization is a hidden tax on every model. How byte-latent transformers attack the $10B cost problem at the architecture level.

Read on Medium → AI Advances

Books

All Books

B*tch Slapped

Raw, practical essays on thriving through life's chaos while building tech careers.

Go Snowflake LLM

Build scalable LLM solutions for startups using Go and modern infrastructure.

Go
LLM
70% Complete Free Chapter

Trendvesting Podcast

All Episodes

Episode 15: AI Trading Strategies

Deep dive into using machine learning for automated trading decisions and risk management.

32 min

Episode 14: Infrastructure Scaling

Lessons learned from scaling Kubernetes infrastructure at high-growth companies.

28 min

Do you actually know what your AI costs?

Let's run a cost & unit-economics teardown — attribute the spend, find the waste, and get a number you can defend to finance.

AI Infrastructure & Unit Economics — weekly

One substantial piece a week on making AI infra legible and its costs attributable: real invoices, cost-attribution patterns, and production teardowns.

No spam. Unsubscribe anytime. Used to send your kit and occasional AI insights.