The business of technology

What’s worth building?

I’m Tera Earlywine, a data engineer and independent consultant. I write about what a warehouse bill can tell you, where an AI workflow needs a person, and how to decide whether a new tool earns its keep.

Exploded layers of ultramarine paper surround a single orange cube.
Inspect the layers

Subtract posted refunds from completed sales.

Migration programs I’ve worked on spanned

500+ workflows

My optimization work identified

$500K+ in projected Snowflake savings

Where do you want to start?

For teams

You’re weighing an AI pilot, a growing warehouse bill, or a migration. Start with one decision and work out what would make the investment worthwhile.

Explore consulting

For technical professionals

You know how the system works. How do you explain the difference a change will make? Try it with a dashboard, a refund, and a short decision note. For a workflow nearing release, use the readiness scorecard.

Try the exercise

Selected work

Decisions behind the build

Staging proof of concept

Could a reviewer retrace an agent’s work?

I designed an audit-agent proof of concept for a client. Recording each step left a trail a reviewer could follow and surfaced gaps in the original process. The handoff documented what still needed work before production.

Inside the audit workflow

Independent build · Active development

What can I hand off when I build with AI?

I’m developing Engine Sandbox to provision and deploy web apps. I use AI-assisted implementation while I own the architecture, requirements, review, and troubleshooting. A useful handoff starts with deciding how I’ll know the work is done.

How I check in on an agent

Prior work · Data platforms

When is a cheaper query worth the engineering time?

My optimization work identified $500K+ in projected Snowflake savings. An estimate gives a team a reason to investigate. Measuring the workload before and after a change tells them what it actually saved.

When SQL optimization pays

Prior roles · Block & Mercari

What does a sales number actually count?

At Mercari I worked in business intelligence and warehouse engineering. At Block I helped set technical direction for financial data systems. I made a small example you can try here. Change one refund and see exactly why the reported total moves.

Trace a refund through net sales

Try it · Illustrative data

What goes into
net sales?

A customer gets $40 back. What should the sales report show? Toggle the refund, then follow the number from the transaction to the ledger.

Reported net sales$560

Business rule

$600 − $40 = $560

Both sales are complete. The posted refund reduces net sales by $40.

Engineering note

Use the refund’s posting date and count each refund ID once. Both sales and the optional refund post in the same reporting period here. Reconcile the result to the ledger for that period.

How I work with teams

Start with the decision.

If a dashboard updated every five minutes instead of every hour, what would someone do differently? I like to ask that kind of question before a team spends money changing a system.

  1. 01Diagnose

    Sit with the person doing the work. Find the delay, mistake, or expense they would change.

  2. 02Blueprint

    Compare the cost of a change with the cost of keeping things as they are.

  3. 03Prove

    Try one complete workflow. Agree on the result that would justify doing more.

  4. 04Scale

    Expand the parts that worked. Identify who reviews exceptions and pays the running costs.

  5. 05Operate

    Give the people running it the checks and instructions they need on an ordinary Tuesday.

About & experience

From the dashboard to the data behind it.

I started in business intelligence at Mercari, writing SQL for people who needed answers. Over 9+ years in marketplaces, fintech, and consulting, I moved from building reports to designing the systems behind them. These days I bring the same curiosity to AI workflows.

Hiring for a team? I’m open to applied AI engineering and senior data roles. Tell me about the role.

2026 — Present

Independent Data & AI Consultant

Tera Earlywine Consulting

I advise engineering teams on platform reliability, modernization, compliance workflows, and applied AI.

2025 — 2026

Staff Data Engineer

Block.xyz

I built and scaled financial data infrastructure and helped set technical direction across platform work.

2024 — 2025

Senior Data Consultant (Qbiz)

Block.xyz

2023 — 2024

Data Engineer

Mercari

2022 — 2023

Data Warehouse Engineer

Mercari

2021 — 2022

Data Analytics Engineer

This is Alice

2020 — 2021

Business Intelligence Analyst

Mercari

2018 — 2020

Jr. Business Intelligence Analyst

Mercari

I developed my foundation in SQL, data modeling, and analytics.

Straight answers

A few things you may be wondering.

Who is Tera Earlywine?

I’m a data engineer, writer, and independent consultant with 9+ years of experience across fintech, marketplaces, and enterprise data platforms.

What can I learn here?

I write about what technology costs, how people use AI at work, and what I learn building with it. Try the net-sales example to see how one refund changes a number that finance and engineering both rely on.

When should a company hire Tera?

Bring me a decision your team is stuck on. I can help you weigh an AI pilot, investigate a warehouse bill, or decide when a migration is ready for its next cutover.

How does an engagement start?

We start with a 20–30 minute fit call. Bring one example and the question you need to answer. If I can help, we’ll agree on the scope and what you’ll take away.

Get in touch

What are you working on?

For consulting, tell me which decision your team needs to make. For a role, talk, or collaboration, choose “Help me scope it” and tell me who the work is for.

Potential engagement

Please do not include passwords or other sensitive information.

Submissions are delivered to Tera by email and stored in Linear for follow-up.

Or book a call directly

Book a fit call