Open to Staff / Senior SDET roles · Bay Area

Sev Rudakov

Staff Software Development Engineer in Test

I build the systems that help teams know when software is ready to ship. Test frameworks, CI pipelines, and the tooling that connects everything together.

I have spent nine years doing this across products with 250M+ users and regulated medical devices, where a missed defect can mean a recall instead of a simple rollback.

A big part of my work is making failures easy to find, easy to understand, and reliable enough that teams can act on them with confidence.

250M+ users on a product I tested ~70% manual regression reduction Primary: Python + Java US work authorized, no sponsorship needed MS Computer Science
// what I actually build

A release pipeline, end to end

Most people describe test automation. This is what it looks like when it runs: unit through mobile, gated on a performance budget, with a report a release manager can read.

jenkins › release-candidate › #1482 idle
pass rate duration tests

// Illustrative run. The stages, stack and numbers come from real work — the log itself is a simulation, not a live build.

// digital twin

Ask my AI twin

It knows my background and answers as me. Every reply is checked by a second model before it reaches you — you will see that step happen. Same idea as a review gate in a pipeline, applied to an agent.

sev@twin — ask me about my work ready
>

An AI twin, not me. It can be wrong, and it will say so when it does not know. For anything that matters, email me.

// impact

Numbers from shipped work

Measured outcomes from production programs, not estimates.

~70%
manual regression reduction
// hours of human clicking, deleted
~30%
post-release defect reduction
// caught before customers did
95%+
CI pass rate
// a green build people trust
50%
coverage increase
// on the paths that actually break
60%
regression cycle reduction
// 6 hours to under 2 hours
250M+
users on the product tested
// consumer scale, Evernote
// experience

Nine years, four companies

Enterprise SaaS, consumer scale, then regulated medical devices — each step raised the cost of being wrong.

May 2023 — Present

Staff Software Development Engineer in Test

Abbott Diabetes Care

Libre glucose-monitoring ecosystem: mobile SDKs, backend APIs and hardware integration under medical-device validation. Framework architecture, CI/CD, and automation standards other teams adopt. Grew suite coverage from ~70% to 90% and cut a regression cycle from 6 hours to under 2 hours.

PythonJavapytest Robot FrameworkAppiumJenkins regulated validation
Jun 2021 — Mar 2023

Senior Software Development Engineer in Test

Evernote

UI, API, web and mobile automation for a consumer product at 250M+ users. Test design and CI integration where a bad release is visible within minutes.

JavaPythonSelenium AppiumTestNGCI/CD
Oct 2018 — Jun 2021

Senior QA Automation Engineer

insightsoftware

Enterprise SaaS UI and API automation. Selenium Grid, cross-browser coverage and CI pipelines across a multi-product portfolio.

Selenium GridJavaREST JenkinsSQL
May 2017 — Sep 2018

Software Engineer, Tools & Automation

Castlight Health

Web smoke, regression and acceptance suites across browsers and platforms — the first place automation replaced a manual release checklist.

SeleniumRubycross-browser
// stack

Tools I use in production

Listed only where the work is real. Nothing here is a weekend tutorial.

Languages

PythonJavaJavaScript SQLBash

Test automation

pytestRobot FrameworkPlaywright Selenium WebDriver / GridAppium (XCUITest / UIAutomator2 drivers) CypressTestNGJUnit GTestPage Object Model

CI/CD & cloud

JenkinsGitHub ActionsGitLab CI CircleCIDockerAWSGit

API & data

Python requestsREST AssuredOkHttp PostmanJSON SchemaOpenAPI FlatBuffers

Devices & grids

BrowserStackSauce LabsXcode Android SDK

Databases

MySQLPostgreSQLOracle

Reporting & tracking

AllureExtent ReportsDatadog JiraTestRail

AI in the workflow

ClaudeCodex test-case generationfailure analysiscode review
// how I work

Opinions, earned the hard way

The short version of what an interview with me would cover.

01

A test suite is a product

It has users — the engineers who read its output at 2am. If the failure message does not tell them what broke and where, the test failed even when it passed.

02

Flaky tests are bugs, not weather

Auto-retry hides the signal and teaches the team to ignore red. Flakes get triaged like defects, with an owner and a root cause.

03

Fast feedback beats total coverage

A six-hour suite gets skipped before a deadline. Cutting one to under two hours changed how often it ran, which is what actually caught defects.

04

Standards travel further than heroics

The goal is a framework other teams adopt without me in the room. Cross-team automation standards outlive any suite I personally write.

05

Regulated work raises the bar

Medical-device validation means traceability, evidence and audit trails — not just a green checkmark. It is a habit I keep even where it is not required.

06

AI belongs in the loop, not in the driver's seat

Claude and Codex speed up test-case generation, failure analysis and review. A human still signs off. Nothing runs unattended in CI on my watch.

// public code

Something you can actually read

Most of my work sits behind an NDA. This one does not.

hackernews-api-test

A Python/pytest API test framework against a live public API: JSON Schema validation on every response, plus a Locust performance gate that fails the build on p95 latency or error rate. Runs on every push in GitHub Actions. Small on purpose — it shows how I structure a suite, not how many files I can produce.

PythonpytestJSON Schema LocustGitHub Actions
View on GitHub
// for recruiters

Answers before you ask

Everything that usually takes three emails to establish.

Where are you based, and do you need sponsorship?

Santa Clara, CA. Authorized to work in the US — no sponsorship needed, now or later.

What roles are you looking at?

Staff or Senior SDET, test infrastructure, and developer-productivity roles where the framework and CI pipeline are part of the job rather than an afterthought. Python or Java shops both work.

Python or Java?

Both are primary. Python for API, service and tooling work with pytest and Robot Framework; Java for UI and mobile suites with TestNG, JUnit and REST Assured. JavaScript, SQL and Bash are working languages, not headline ones.

Do you write real code, or only tests?

Frameworks, internal tooling, CI infrastructure and API/SDK integration alongside the suites. At Abbott that includes unit tests for firmware-adjacent components in C++ and limited native work on an in-house integration framework touching Swift and Kotlin.

How much of this is mobile?

A large share. Appium with the XCUITest and UIAutomator2 drivers, on real devices and on BrowserStack and Sauce Labs, across both a consumer app and a medical-device companion app.

What is your experience with AI tooling?

Daily, as an assistant: Claude and Codex for test-case generation, failure analysis and code review, with a human reviewing every result. I have not run agents unattended in CI or owned an org-wide AI testing strategy — if that is the job, ask me and I will tell you exactly where the line is.

What does regulated medical-device work actually mean here?

About three years of it, at Abbott only. Validation evidence, traceability from requirement to test to result, and controlled change — on top of normal engineering. The rest of the nine years is consumer and enterprise software.

Fastest way to reach you?

Email. I answer every message that names the company, the level and the base range.

// contact

Let's talk about your release process

If your team ships nervously, that is a solvable problem. Tell me what breaks and I will tell you how I would approach it.