Overcommitted | Software Engineering and Programming Insights

Roxy Rodriguez-Becker taught middle school math, moved into the school library after earning a master's in library and information science, and then taught herself to code while raising three kids. She is now a developer community manager at Torc and the host of the She Ships podcast. In this episode she walks through what that transition actually required: the four AM club, waking up to study until she rushed out the door to a five AM gym session, working through 100Devs, and treating the job search as a numbers game with fifty to sixty outreach messages a week. She is candid about the burnout that followed and what it took to come back from it.The conversation also digs into networking as a skill rather than a personality trait. Roxy describes going to her first conference and barely speaking, then meeting her future colleagues at the second one, and shares tactics that work for socially awkward people. We also talk about the Tube Feed Tracker, the web app she built after watching ICU nurses struggle with feed rate math for her medically complex son. It grew to roughly four thousand users and now includes BMI based rate suggestions and a calorie tracker covering seventy verified formulas. Roxy closes with why career switchers cannot go it alone, and why community matters more than any single course. Plus a tiny internet troll lightning round covering course playback speeds, salmon shirtgate, and conference karaoke.

Hosts
Overcommitted: https://overcommitted.dev
Bethany Janos: https://www.trustyduck.dev
Brittany Ellich: https://brittanyellich.com

Guest
Roxy Rodriguez-Becker: https://linktr.ee/roxyrodbeck
She Ships on YouTube: https://www.youtube.com/@roxyrodbeck
Roxy's blog: https://foxyroxy.dev/
Roxy on Bluesky: https://bsky.app/profile/roxyrodbeck.bsky.social
Roxy on X: https://twitter.com/RoxanaCodes
Roxy on LinkedIn: https://www.linkedin.com/in/roxana-rodriguez-becker

Links
Tube Feed Tracker: https://tubefeedrate.com/
About Tube Feed Tracker: https://tubefeedtracker-about.vercel.app
Torc: https://www.torc.dev/
100Devs: https://100devs.org/
Scrimba, Learn JavaScript: https://scrimba.com/learn-javascript-c0v

Closed That Tab is back. Erika finally read Richard Cook's How Complex Systems Fail, an 18 point paper Cook wrote with patient safety and medical systems in mind, but one that is frequently discussed in web development and software engineering since the foundational principles apply universally. Erika and Bethany work through what holds up in practice.

They cover the difference between normal failure and catastrophic failure, why knowing your system's baseline is its own skill, and why fixes are never free: every change adds complexity, knowledge gaps, and new risk. They get into safety as a set of defensive actions, including rate limiting, caching, rollback plans, feature flags, backups, playbooks, and watching for drift. Then they take on Cook's most contested claim, that root cause analysis is fundamentally wrong for complex systems, and where they agree and disagree. Bethany makes the case that incident reviews are still one of the best learning tools available, as long as psychological safety comes first.

The conversation also digs into hindsight bias and self-blame, the difference between taking accountability and taking blame, practitioners as both the risk and the defense in a system, why training and safety work are chronically undervalued, where AI helps during an incident and where you still need a human expert, practicing failure through game days and shadowing, and keeping on-call sustainable so you do not burn out the people holding the pager.

Hosts
Overcommitted: https://overcommitted.dev
Bethany Janos: https://www.trustyduck.dev
Erika Eggemeyer: https://github.com/eggyhead

Links
How Complex Systems Fail by Richard Cook: https://www.adaptivecapacitylabs.com/HowComplexSystemsFail.pdf
Thinking In Systems: https://www.chelseagreen.com/product/thinking-in-systems/

Brian Okken, lead engineer at Rohde & Schwarz, creator of the pytest book and course, and author of the new book Lean TDD, joins us to rethink test driven development. His core argument: write tests at the highest level that still makes sense, so you can refactor, restructure, and hand work to an agent without rewriting your suite every time. We dig into why function-level tests with heavy mocking are often harder to write than system-level tests, why the "TDD takes too long" complaint usually comes from testing at the wrong level, and why old rules about test speed no longer hold up on modern machines and CI.
The conversation also covers the lean side of Lean TDD: what counts as waste in a test suite, when to delete tests you wrote on the way to a better one, and why a thousand tests is not automatically better than twenty. Brian shares how caring about the whole system rather than just his slice of it changed his career, how he works with AI agents while still reading every line they produce, the stigma around test engineering roles, and what he learned self-publishing a 100 page book and recording the audiobook himself. He closes with advice for newer engineers on large codebases and why everyone is winging it more than you think.

Lean TDD Book: https://leantdd.com/

Free sample chapter (chapter 1): https://leantdd.com/downloads/lean-tdd-chapter-1-sample.pdf

Brian’s Blog: https://pythontest.com/

Brian’s LinkedIn: https://www.linkedin.com/in/okken/

Brian on GitHub: https://github.com/okken

pytest course: https://pythontest.com/courses

Test & Code podcasthttps://pythontest.com/testandcode

Hosts
Overcommitted: https://overcommitted.dev
Brittany Ellich: https://brittanyellich.com
Erika Eggemeyer: https://github.com/eggyhead

SummaryBrittany and Erika close out the Overcommitted book club's read of Tanya Reilly's The Staff Engineer's Path with guest Jade Dickinson, a senior software engineer at Henry Schein One. They dig into glue work and how to make it visible for promotion, the book's staff engineer archetypes, the "why have we stopped" blockers chapter, and what it means to be a delegator on a staffed project. Jade also talks about her path into software after working in GIS mapping for Ebola response, and the group debates whether AI is shifting the scope senior engineers are expected to own. It closes with a staff engineer lightning round: Elm or React, glue work as underrated or a trap, and what a staff engineer is actually for.Links
Jade's Blog: https://jadedickinson.com/
Jade on Bluesky: https://bsky.app/profile/jadedickinson.bsky.social
Jade on Mastodon: https://ruby.social/@jadedickinson
Helvetic Ruby Coneference 2026: https://helvetic-ruby.ch/talks/
Jade's React to Elm Talk: https://www.youtube.com/watch?v=_CdHa4_g3EM&list=PLhBlV3WSUrT3jCmqmYWcUEvmon_6K6WUy&index=18
Staff Engineer’s Path: https://www.noidea.dog/staff
Glue work: https://www.noidea.dog/glue
Open and Async book: https://open-and-async.com/
Staff Engineer Archetypes by Will Larson: https://staffeng.com/guides/staff-archetypes/
How I ship post from Sean Goedecke: https://www.seangoedecke.com/how-to-ship/
Hosts
Overcommitted: https://overcommitted.dev
Brittany Ellich: https://brittanyellich.com
Erika Eggemeyer: https://github.com/eggyhead

SummaryIt's a hosts-only week. Brittany and Erika catch up on what they're learning (Erika is starting Designing Data-Intensive Applications; Brittany just began Georgia Tech's online CS master's with a computer vision course) and then work through the week's AI news.First up: OpenAI's GPT-6 Astra, billed as the company's most aligned model. Brittany and Erika question what "aligned" even means when the headline numbers are a 100% on ExploitBench and a 0% honeypot-cheating rate, and whether a model that's harder to monitor and scores perfectly is aligned or just good at taking the test. That leads into how each of them actually evaluates a new model (spoiler: vibes), why Brittany has drifted back toward cheaper models she understands, and whether frontier models are now a research product rather than a consumer one.They also revisit the AI 2027 paper from their episode last year and how many of its predictions have already landed, debate whether GitHub Copilot approving pull requests is a real workflow change or just signaling, and talk about why the arrival of open-weight models with serious exploit capability means now is the time to get serious about security basics.Brittany closes with ATProto Spaces, the new alpha that brings non-public data to atproto, what it enables for groups and communities, and how her work at Bluesky connects to it.Links
OpenAI Release: ⁠https://openai.com/index/gpt-6-astra/⁠
AI 2027 Episode: https://overcommitted.dev/ep-13-ai-2027-will-ai-take-my-job/⁠
AI 2027: https://ai-2027.com/⁠
Copilot Approving PRs: https://github.blog/changelog/2026-09-01-copilot-code-review-can-now-approve-pull-requests/⁠
Atproto Spaces Release: ⁠https://atproto.com/blog/atproto-spaces-alpha⁠Hosts
Overcommitted: ⁠⁠⁠https://overcommitted.dev⁠⁠ 
Brittany Ellich: ⁠⁠https://brittanyellich.com⁠
Erika Eggemeyer: ⁠⁠⁠https://github.com/eggyhead⁠