Virtual Coder

A consultancy run by people who still ship.

Virtual Coder exists because the best technical advice comes from those carrying the consequences of it.

Virtual Coder is led by Ish Sethi, currently Chief Technology Officer at Pluk, where he leads engineering for a research platform that evaluates people and companies at scale.

Before Pluk came four years at Microsoft, building and operating Microsoft Teams at a scale where an edge case is a support incident for millions of people. Before that, a venture of his own: Simplified Automation, co-founded in 2018 and led as CTO until the market closed it in 2021. Both experiences inform the advice — one taught what operating at scale actually costs, the other what it costs to be wrong.

The firm keeps building its own systems for the same reason. The work in the portfolio runs in production, with its own migrations, its own on-call and its own users. Every estimate we give is calibrated against that rather than against a spreadsheet.

We take a small number of clients at a time, deliberately.

Background

Where the judgment comes from.

  1. Present

    Chief Technology Officer, Pluk

    Leading engineering for a research platform that evaluates people and companies at scale.

  2. 2021–2025

    Software Engineer, Microsoft Teams

    Four years building and operating one of the largest collaboration products in the world.

  3. 2018–2021

    Co-founder and CTO, Simplified Automation

    Built and scaled a venture from nothing, and closed it when the market moved. The most expensive lesson available, and the most useful.

Principles

Four positions we bring to an engagement.

An agent is an interface, not a product

A tool that bypasses your service layer is a second copy of your business rules that nobody is testing. Agent tools call the code your application already trusts, or they do not ship.

Estimates should come from operators

We keep our own systems running in production, so our estimates account for the migration, the authentication edge case and the second week after launch. That is where the time goes.

Small increments beat large plans

Two weeks is long enough to deliver something real and short enough that being wrong is inexpensive. You should be able to stop after any increment and keep what was built.

Write the decision down

Architecture calls, trade-offs and the options deliberately rejected are documented. It is the difference between a system you own and one you have inherited.

Tell us what you are building.

A thirty-minute scoping call, and a straight answer on whether we are the right firm for it.