Cubic Pixel is one person. There are no vacancies, no hiring pipeline and no recruitment budget, and a product built on not overstating things is not going to run a decorative jobs board. What follows is what this work is actually like, and what would make a speculative message worth reading.
Tessira is free to use and has no billing in it, so there is no revenue to hire against. The studio has no employees, no office and no outside investment. Nobody is being interviewed, no roles are being drafted, and there is no shortlist somebody is already on.
If that changes, this page changes with it — the vacancy will be described here in specific terms, with what it pays, before anyone is asked to spend an afternoon writing an application. Until then, treat this page as a description of the work rather than an invitation to apply.
Not values chosen for a careers page — these are the working rules that already govern every change that ships, and anyone joining would inherit them.
A change is not finished when the code reads correctly. It is finished when the page has been rendered in a real browser, measured and looked at. Nearly every serious bug here was caught that way, and nearly every one that escaped was reasoned about instead.
Marketing copy is held to the same standard as code. If the app cannot do it, the page does not claim it — which is why you will find “encrypted at rest” here and never “end-to-end encrypted”, and why an external AI provider is named rather than implied.
Guarantees that live in a policy document are decorations. The ones that matter fail closed in code: the checks that look for identifying data in an outbound request throw rather than transmit, and account isolation is proved with adversarial tests rather than asserted.
Whether a check mark is optically centred, whether a section heading repeats correctly across a page break, whether an animation loops when it should stop after three passes. Users cannot articulate these things, but they can feel all of them.
Decisions, constraints and the reasons behind them are recorded next to the code, in the same change that makes them true. A one-person project only survives because the notes from six months ago are still accurate.
Every feature is a maintenance obligation for one person, indefinitely. The default answer to a good idea is no, and the ones that get through have to displace something or justify their existence for years.
Written to put people off as much as to attract them. If none of this sounds appealing, that is useful information for both of us.
One edge worker, a key-value store, static assets and a headless-browser rendering binding. The interface is plain HTML, CSS and JavaScript with no framework, no bundler and no build step — which is liberating if you actually know the platform, and uncomfortable if your instinct is to reach for a library.
You would need to be fine with reading a specification rather than a package's documentation, and with the fact that “add a dependency” is usually the wrong answer.
A significant part of this product is typesetting for A4 paper through a browser engine. Page breaks, repeating headings, margins that a browser will not paint, fonts that must be embedded because the render environment cannot reach the network. The rules are unintuitive and mostly undocumented.
The way you learn them is by rendering, rasterising and looking at the output — repeatedly, and often to discover that the obvious approach is empirically impossible.
The product is free, so every rendered PDF and every AI request is a real cost carried by one person. That is why results are cached against their content, why an idle browser session is never left running, and why nothing analyses your CV unless you clicked something.
Engineering decisions here have prices attached, and you would be expected to think about them rather than treat capacity as infinite.
The content in this app is somebody's employment history, and losing it or leaking it is not a bug you recover from with an apology. Data handling gets treated with more caution than the feature it serves, and “it probably works” is not an acceptable standard for anything touching storage, encryption or the AI boundary.
The same person designs the page, writes the code, writes the copy on it, deploys it and answers the support email when it breaks. There is nobody to hand a half-finished thing to, and there is no design team to reconcile with. The upside is that nothing is lost in translation; the downside is that you own it entirely.
These are not vacancies and nothing here is being recruited for. They are an honest description of where a second pair of hands would make the most difference, so a speculative message can be aimed at something real.
Perfectly welcome, as long as you know what it is: a note to one person who currently has nothing to offer you. It will be read, and you will get a reply, and that reply will probably say there is still no role.
If you want it to be worth your time, four things make the difference:
A product about honest hiring documents has no business being careless with the people who write them.
No “small paid-in-exposure project”, no speculative template designs, no take-home exercise that quietly produces something shippable. If work is done for this product, it is paid for.
Every message that is clearly written by a person to this studio gets a reply, including the ones where the answer is no. Silence as a rejection is cowardice with a diary excuse.
No listings kept open to collect candidates, no role advertised before it is funded, and no salary range invented to look competitive. When there is nothing, this page says nothing — as it does today.
There is no candidate database, no applicant tracking system and no talent pool — an email that arrives is just an email in an inbox. That is the honest answer, and it means the practical thing is to write again if a role is ever posted here.
If you would rather your message was not retained at all, say so in it and it will be deleted once it has been answered.
Not currently. Supervising someone well is a substantial time commitment, and doing it badly is worse than not doing it — an unsupervised intern on a product handling people's personal data is a bad idea for everyone involved.
Not at present. If a specific piece of specialist work ever needs commissioning — an accessibility audit, a security review, a template designed by someone better at typesetting than I am — it would be a paid, scoped engagement, and it would be mentioned here.
No. Tessira exists partly because nobody is being paid to move a metric, and taking money to grow would change what gets built almost immediately. That is not a moral position about funding in general, just a description of what this particular product is for.
Yes, and hello@cubicpixel.co is the address for it. Questions will be answered accurately, including the unflattering ones — what the product does not do is on the about page in as much detail as what it does.
Then you are in the right place for the wrong page. The CV writing guide and interview tips are written for you, and the app itself is free — no card, no trial, no watermark on what you download.
The most useful thing this page can do today is send you back to the product it describes. Start a CV, and tell us where it falls short.