Software studio · est. 2019

We build the parts of software no one notices until they break.

Zhendao is a 34-person engineering studio that designs platform infrastructure, internal tooling, and reliability systems for teams whose outages make the news.

40.7N · UPTIME 99.982% FIG. 01 — SIGNAL, NOT NOISE
zhèn — quake
dào — the way

The name comes from a seismograph, not a slogan. Zhèn is the tremor — the load spike, the failed deploy, the 2 a.m. page. Dào is the path a well-built system draws through it: absorbed, logged, recovered, boring.

Founder Wéi Zhao started the studio in Shenzhen in 2019 after six years running infrastructure at a payments company that treated downtime as a marketing problem instead of an engineering one. Zhendao takes the opposite position: the interface is not the product. The thing underneath it is.

We're a studio of 34, split between Austin and Shenzhen, working with roughly nine clients at any given time. We turn away more projects than we take — mostly ones where the ask is a redesign and the actual problem is a database that can't hold the load.

What we actually get hired to fix

01 — 04
01

Platform engineering

Internal platforms that let product teams ship without filing a ticket to the infrastructure group. Golden paths, not gatekeeping.

KubernetesTerraformInternal PaaS
02

Reliability engineering

SLOs that mean something, on-call rotations people can survive, and postmortems that change the system instead of assigning blame.

SREIncident responseObservability
03

Data-path migrations

Moving a company off a database, queue, or monolith it has outgrown, without a weekend of downtime or a resume-generating event.

PostgresKafkaZero-downtime cutover
04

Developer tooling

CLIs, CI pipelines, and internal dashboards built for the fifty engineers who use them daily, not the one VP who sees the demo.

CI/CDDXInternal tools

A few things we've shipped

SELECTED, NOT COMPREHENSIVE
Freight & logistics — platform rebuild
Kiln
Replaced a Jenkins farm held together by cron jobs with a deploy pipeline a 41-person engineering org could actually trust.
14m → 96s
median deploy time
−73%
failed rollouts, 6mo
Consumer fintech — data migration
Ledgerline
Moved 2.3B ledger rows off a sharded MySQL setup onto Postgres with logical replication, live, during business hours.
0s
customer-facing downtime
2.3B
rows migrated
B2B SaaS — observability & on-call
Fathom Rotations
Rebuilt alerting from scratch after a client's on-call engineers were paging out at 4x industry norms. Cut noise before adding more dashboards.
312 → 47
pages / week
19min
avg. time-to-ack, down from 51

How an engagement runs

TYP. 8 — 14 WEEKS
i.

Audit

Two weeks embedded with your team, reading logs and runbooks before we write a proposal, not after.

ii.

Scope

A written plan with the failure modes we're solving for named explicitly, so we're not graded on vibes.

iii.

Build

Weekly demos against a real environment. No slide decks standing in for working software.

iv.

Hand-off

Runbooks, recorded walkthroughs, and 60 days of on-call shadowing before we step back.

Who you'd actually work with

3 OF 34

Wéi Zhao

Founder & Principal Engineer

Six years running payments infrastructure before starting Zhendao. Still writes code on client projects most weeks.

Priya Nandakumar

Head of Platform

Led the Ledgerline migration end to end. Previously spent four years on a database internals team.

Tomás Ferreira

Head of Delivery

Runs client engagements and refuses to let scope creep in without a written change order.

Tell us what keeps paging you at 2 a.m.

AUSTIN, TX · SHENZHEN, GD
REPLIES WITHIN 2 BUSINESS DAYS
CURRENTLY BOOKING FOR Q1 2027