Hi, my name is Radon Rosborough (they/them). You may know me as @raxod502 online due to a juvenile teenage username choice that is now a load-bearing component of some critical open-source infrastructure and therefore can never be changed.

I live in San Francisco, CA.

Professional work

Since 2020, I have worked full-time professionally as a software, infrastructure, and networking engineer. This includes things like:

  • Container orchestration and service deployment frameworks
  • Zero-downtime maintenance and replacement of legacy systems
  • Internal tooling, system documentation, and automation
  • Low-level operating system and network debugging
  • System monitoring, automated alerting, and incident response
  • API reverse engineering and security hardening
  • Open-source project development, contribution, and community management

If you would like to recruit me, please read contacting me, which lists several sectors I refuse to work in. Unsolicited job offers for those areas will be reported as spam.

Skills

In interviewing over 70 candidates for engineering positions, I have learned that a resume tells you almost nothing about whether someone has any relevant technical skills. Instead of a resume, I offer examples of my publicly available work and writing for your evaluation. It is my hope that this will be more convincing than a resume to the kind of people with whom I prefer to work.

Past experience

From 2020 to 2025, I worked at Plaid on the Infrastructure team of about ten engineers. Though the specifics of ownership changed over time, Infrastructure was responsible for around seventy different services in all, including Plaid’s Kubernetes clusters, microservice framework, deployment systems, continuous integration, network topology and service mesh architecture, package management, PKI, infrastructure-as-code, cost management, databases, logging, disaster recovery, metrics aggregation, alerting pipelines, tracing, on-call tooling, and many other individual components. Although I worked on everything the team owned at one time or another, my focus (by choice) tended to be on improving operational maturity: for example, making pager alerts more actionable and less frequent; transforming ad-hoc procedures by individuals into documented steps that can be used and improved by anyone on the team; and correcting structural issues such as nebulous ownership boundaries which tend to create technical debt over time.

For ethical reasons, I changed jobs in 2025 to instead work as an infrastructure and reverse engineer at Beeper. Large tech companies like Apple, Meta, and Google perpetuate horrific abuses to people and society, and I feel it’s important to reduce the de-facto monopoly power they exercise at the expense of consumer rights and human decency. Given that the network effects of closed chat networks are one very effective way that these companies maintain their power, I feel that my work at Beeper to make those networks interoperable by means of open-source Matrix bridges (that people can run themselves if they wish) is a specific way I can contribute to positive social change and that I can give people more power against the corporations that want to abuse them.

Personal background

I grew up in Boulder, CO, went to college for computer science at Harvey Mudd College in Claremont, CA, and subsequently settled in the San Francisco Bay Area where I intend to stay for the foreseeable future.

I am queer, transgender, and non-binary, and use they/them pronouns. I am also on the autism spectrum, and experience anxiety and depression as well as chronic pain. Finally, I am white, economically well-off, and a US citizen. These aspects of my identity and experience, as well as what I have learned from the people around me, have informed my worldview, which is socially and politically progressive.

Values and politics

While every person’s value system is too complex and multidimensional to fully capture in writing, I nevertheless offer here a few specific examples of beliefs that I hold on contested topics.

Most contemporary software design principles are catastrophically anti-human. The majority of consumer software (webapps, mobile apps, smart devices) is terribly designed, and paradoxically high-budget companies like Apple, Facebook, and Google often use those resources in ways that make their products worse rather than better. There is a bizarrely celebrated trend towards infantilization in consumer software design, where it’s taken for granted that complexity is irreducible and impossible to manage, and therefore the only way to produce a superior user experience is to remove features, stop people from making their own decisions, and raze everything that strays off a single golden user story. I believe in technology that teaches people implicit skills as they use it, and empowers them to understand, control, and achieve more over time. People should use software as a tool, not the other way around, and the idea that manipulating a computer is somehow an inaccessible or specialized skill is a lie peddled by companies with an interest against a skilled and engaged community.

Security is best achieved through community solidarity, not restriction of freedom. The word “security” as used by both tech companies and national governments is so broad as to be essentially meaningless, and justifies wildly inappropriate behavior in both cases. It is of paramount importance to ask: “security for whom? security from whom? and security to what end?” Militaristic solutions rarely produce meaningful security in any useful sense, especially in the long term. For example, I feel day-to-day safety in a city where I feel connected to my neighbors and where I know that everyone has access to the resources they need to survive, not in one where potential “suspects” of misconduct are constantly being shaken down by police. In the realm of technology, I feel comfortable in my digital privacy when I use open-source software packaged by a network of cooperative volunteers, not when every app I install has to be personally approved by Big Brother Google. The fact that open-source software is as or more secure and trustworthy compared to proprietary software, despite receiving billions of dollars less in funding and venture capital, is an astonishing indicator of how much better a job can be done by community-oriented security solutions.

Efficiency and commerce are not fundamental social goods. Most tech companies (and especially startups) are making the world a worse place. Even setting aside the ones that participate in straightforwardly evil monopoly practices, abuse of employees, or support of right-wing extremism, there is a baseline way in which tech companies tend to exacerbate income inequality and divert resources from existing communities towards shiny but ultimately ineffective technologies. As such, an ethical company needs to actively produce social good. However, even taking this premise for granted, many people seem to have a lax understanding of social good, and consider simply “disrupting an outdated industry” or “automating a slow process” or “making it easier to build things” to be a self-evident positive good for the world without giving any thought to the actual concrete impact these structural changes will have on the everyday people they affect. Introducing technology does not automatically make people’s lives better, especially when the company introducing it is primarily trying to extract money rather than actually help members of a specific community that they have an understanding of and personal relationship with.

Open-mindedness is not the same thing as gullibility. It behooves me to keep an open mind in the sense that the amount of knowledge in the world greatly exceeds what I can possess at any given time. There are many parts of the world and of other people’s experiences that I know I have insufficient information to make statements about, and sometimes my beliefs turn out to be based on wrong or biased information. However, not every alternative idea should be treated equal: this is the fallacy of false balance, which has been weaponized by many contemporary hate groups to give people and reporters the impression that there is still a legitimate discussion to be had about their long-since-discredited conspiracy theories. It is both foolish and intellectually lazy to pretend that claims like “modern vaccines might be too dangerous to be recommended” and “it might be too easy for trans children to access gender-affirming care” are good-faith suggestions that should be discussed in any way other than to reference the extensive scientific literature that has already ruled them out as plausible ideas.

Most contemporary usage of AI is both anti-intellectual and anti-community. The underlying technology is not inherently evil. However, LLMs seem to attract people with bad ideas like flies to honey, partly because of a very expensive campaign by AI companies to sell lies about what they are and are not capable of, but also because the technology is ideal for appearing to be more sophisticated than it actually is, and for creating products that appear to work better than they actually do. While there are certain small, well-defined tasks that a skilled software engineer can make good use of AI models for, the general trend to try to write code as fast as possible using an LLM is completely uncompelling to me. As a software engineer, I do not simply churn out code. I also engage in an ongoing learning process about each level of the software stacks I work with, identify limitations in open-source components that affect my development and contribute improvements to them, build improved tooling, and use my work experience to develop creative solutions to improve the software communities that I am a part of, informed by my actual on-the-ground experience of configuring, contributing to, documenting, and deploying the software in those communities. Using AI to skip out on all the opportunities to learn about how languages and operating systems work in practice, auto-generating hacky boilerplate code rather than contributing improvements to frameworks or building better ones, and generally taking a hyper-individualist approach to software development that lacks any meaningful community engagement, is just not something that has any appeal whatsoever to me. The problem here isn’t the technology, it’s the mindset. A lot of people were already approaching programming this way, and I always disapproved; the availability of generative AI models has just made it a lot easier to cheap out on giving back to the communities you benefit from.

Contemporary leftist organizing strategy requires dramatic improvement before it can construct a practiceable alternative to neoliberal capitalism. Have a read through Inventing the Future, or the summary in that Wikipedia page.

Personal interests

I like:

Technology choices

People are often curious: if I eschew Apple and Google, then what do I use? I’ve done considerable work to identify alternatives across the hundreds of proprietary services I previously depended on, and some of that work might be the subject of future writing. For now, here is a short list.

Laptop/desktop

  • Laptop: Framework for the repairability and Linux support. The ones with AMD graphics cards, because fuck NVIDIA. I also have a System 76 laptop but find it less robust. For a home server/NAS, I use a cheap ThinkPad X201 from eBay.
  • Desktop operating system: Arch Linux exclusively due to the greatly superior packaging ecosystem and documentation. Anything that can be done on another distribution can be ported to Arch with minimal effort; the converse is not true.
  • Desktop environment: GNOME with some minimal configurations.
  • Filesystem: BTRFS for subvolume and snapshot support, installed in an encrypted LUKS volume (with /boot encryption) accessed by GRUB.
  • Server operating system: either Arch Linux or Ubuntu, the latter for systems that don’t require too much upkeep since it’s more broadly supported out of the box by hosting providers and isn’t too inconvenient for simple use cases.

Other devices

  • Phone: For now, GrapheneOS compiled from source on the Google Pixel 6a, with KernelSU, LSPosed, and ReVanced modification frameworks. Apps installed from F-Droid or Aurora Store. Automation using Tasker and AutoNotification. Google services either sandboxed or via microG; push notifications via ntfy. Keyboard from FUTO. In the future, /e/OS on the Fairphone. Only apps that have a really good reason to have to be apps physically installed on my phone are supported; anything else I either use via web browser or run in the Android emulator (GrapheneOS build) on desktop.
  • Server hosting: PikaPods for off-the-shelf open-source applications; Hetzner for most ad-hoc physical servers; Railway for mostly-idle custom applications and some cron jobs (with Sleeping Beauty to minimize billing); Fly.io for anything beefier or with durability requirements; Amazon EC2 for my Minecraft server hosting, to take advantage of cheap billing for stopped instances, but with a project in the works to move that hosting elsewhere.
  • Application monitoring: Self-hosted Uptime Kuma for uptime monitoring, Netdata Cloud for server metrics, Healthchecks.io (one of my favorite and best-run services) as back-stop and for cron job monitoring, and some custom Node.js code in a Railway container to perform more advanced checks on specific applications such as SMS deliverability.
  • Networking: Router running OpenWrt with ip6neigh; VLAN by Tailscale.
  • Personal device tracking: Self-hosted FMD server.
  • Sex toy control: Buttplug.io.
  • Website analytics: very minimal, via self-hosted Umami instance that collects no tracking information but can still be easily blocked by default uBlock Origin rules if you wish.

Software

Online services

  • Backups: On-site backups use BTRFS snapshots generated by Snapper. Off-site backups are performed by Borg, configured by Borgmatic, and stored on BorgBase. PikaPods backups are stored in Backblaze B2.
  • Document sharing and collaboration: Self-hosted Nextcloud server with Collabora Online.
  • File synchronization: Syncthing installed on my laptops, phone, and a virtual private server for reliability.
  • Flight planning: Skyscanner.
  • Mail provider: Fastmail, unambiguously the best for their excellent standards support and contributions to new standards. I also have a self-hosted mail relay integrated with my Fastmail account (open-source and write-up pending) that filters embedded advertisements out of mail from certain senders who stoop to such depths.
    • Contacts and calendar: Both provided by Fastmail. Synchronized with mobile using Davx⁵.
  • Messaging: Signal, SMS/MMS (with self-hosted Zipline and PrivateBin servers to include high-resolution attachments and disappearing messages), or email. Everyone has access to at least one of these if they’re reachable digitally at all. To synchronize messages across all my devices and handle backups, I use self-hosted Matrix bridges accessed via Beeper and Element. Specifically: beeper/android-sms, mautrix/signal, mautrix/telegram, mautrix/discord.
  • Navigation: For now, HERE WeGo on both desktop and mobile.
  • Newsletters: Where possible, via RSS using rss2email and/or self-hosted RSSHub instance. Otherwise by email subscription.
  • Note-taking and to-do lists: Primarily Emacs Org Mode (with agenda, capture, clock), with Obsidian for mobile note-taking and organice for mobile access to Org files.
  • Online forms: Jotform for now.
  • Online videos: For YouTube content, I track subscriptions with a self-hosted Invidious instance, then download desired videos with yt-dlp and watch them in VLC. In the case of visiting the actual YouTube website, SponsorBlock and Remove YouTube Shorts are mandatory. I also subscribe to Nebula for creators available there.
  • Photo storage: Self-hosted PhotoPrism instance. Photos are uploaded from Android by PhotoSync, or via self-hosted PhotoPrismProxy instance.
  • Reminders: Small Flask server running on a VPS, with a Tasker shortcut to transcribe voice notes to it from my phone. Laptops poll the server for reminders and display them as persistent push notifications.
  • Search engine: Kagi for the automated removal of AI slop and ability to customize search results to further tune quality, or DuckDuckGo for untrusted systems. TinEye for reverse image search.
  • Social networks: None, I maintain relationships in real life and via direct messaging.
  • Translation service: DeepL for now, though I am looking at alternatives.
  • Videoconferencing and calls: Signal, Jitsi, or direct phone call.

Finance

  • Financial institutions: Elevations Credit Union (personal checking, debit, credit, check writing, bill pay) and SF Fire Credit Union (business checking, debit).
  • Money transfer: credit union bill pay (ACH via email or SMS), physical cash, Western Union, Wise, or if absolutely necessary Venmo (with Venmo Auto Cashout). To ease transfer bookkeeping, Splitwise (for now).
  • Donations: Via Liberapay or Open Collective where possible, or direct via ACH transfer or debit/credit. Some donations are done through Patreon or GitHub Sponsors… for now.
  • Accounting: Custom tool, to be open-sourced.

Media/purchases

  • Books: Mostly borrowed from one of eleven different California libraries, either physically or via Libby. Purchases managed by Calibre and read on Kobo Libra H2O.
  • DRM: No statement is made here about my personal interactions with DRM, but it can be observed that libgourou conveniently removes Adobe DRM from ebooks and noDRM/DeDRM_tools removes a number of other DRM formats, including legacy Kindle DRM from ebooks. It can also be observed that The Pirate Bay, Anna’s Archive, and Transmission can be used in combination with a port-forwarding VPN such as AirVPN to obtain copies of media that are not otherwise available for DRM-free purchase anywhere. I strongly recommend making legitimate purchases of media whenever it is possible to do so, to reward authors and publishers for selling DRM-free.
  • Graphic novels and comics: Mostly purchased in hard copy from various San Francisco bookstores.
  • Music: Purchased exclusively DRM-free, mostly from Bandcamp. Music library managed by custom software to be released. For now, also some streaming from YouTube Music in concert with ReVanced.
  • Podcasts: On Pocket Casts.
  • Shopping: From local brick-and-mortar businesses where possible, and otherwise online from wholesale suppliers or other storefronts specialized in the relevant area. Groceries, pharmacy, and household supplies from neighborhood businesses.
  • Video games: Purchased DRM-free from sites like GOG or itch.io where possible. For Steam games, an open-source client is in the works.
  • Webcomics: Tracked via Mihon with keiyoushi/extensions.