Founding Design Engineer | CodeSpeak

Role: Founding Design Engineer (Senior / Staff)

Company: CodeSpeak

Location: London, on-site. Ideally you already live in London. Candidates based elsewhere are welcome to apply, and remote is possible for a very strong candidate, but UK-based candidates come first. Candidates with a Global Talent visa, granted or in process, are a good fit.

Job Type: Full-time

About CodeSpeak

Agents write more code every month. Human brains stay the same. CodeSpeak is building a program model that sits one level above code, so people can see how a program works and changes over time without reading much of it. The goal is to give control back to the developer: you think about architecture, and the routine goes away.

CodeSpeak was founded by Andrey Breslav, the creator of Kotlin, which millions of developers use daily. Around him is a team that has built the tools other engineers rely on, at Google, Databricks, Perplexity, JetBrains, Sourcegraph, and Elastic. When it comes to program analysis and developer tooling, these are some of the best people in the world. CodeSpeak is backed by Accel and by funds that also invested in Hugging Face and Mistral, and funding covers several years.

Here is the open problem. The team has a fair idea of what the program model should look like. It does not yet know how to visualise it. As Founding Design Engineer you will decide how it looks, then build it and ship it. The core engineering team has just been hired, and nobody on it has design-engineering skills. You are the missing piece.

CodeSpeak is a UK company. It has no business in Russia and no ties to Russian companies. It is still testing which problems it can solve best, so priorities may shift.

What you are expected to do:

Design and delivery

• Own how people interact with CodeSpeak: in the CLI, on the web and on any future platform.

• Design the interfaces yourself. Build many prototypes, decide on the right UX and ship the result.

• Harden the prototypes users liked, so they hold up on real data, with the same care for performance and product quality as for design.

• Ship features to production without help from the rest of the team.

Product problems

CodeSpeak works with unusually complex information: developer intent, requirements, code, architecture, the relationships between them, how they change over time, contradictions and dependencies. The job is not to display it, but to help developers understand it, navigate it and act on it.

• Find a way to visualise CodeSpeak's new, higher-level program model.

• In your first 30 days, visualise the architecture of a codebase split into components. Go deeper from there: a single component, its diff and its code review.

• Then interfaces that let developers navigate structured requirements and how they relate to each other, review requirements alongside code review, see how requirements map onto parts of the codebase, explore how intent, architecture and code change over time, browse agent sessions, find contradictions between requirements, and edit and reorganise requirements rather than only view them.

• Web UI is the simplest and most obvious platform. Some of this will live in the console. VS Code plugins and other desktop frameworks may follow, and the final scope is still open.

Ways of working

• Find tasks and clarify requirements without waiting for a brief. Nobody will micromanage you. Egor and Andrey give general direction.

• Work directly with the founder and the engineering team to turn early ideas and hypotheses into intuitive, interactive product experiences.

• Take part in design meetings, where many decisions are made on the spot.

• Explain and defend your design decisions to Egor and Andrey. Both hold strong opinions, and both argue hard.

• Argue for your tasks, and step back with good grace if the team disagrees.
  

What you need to succeed:

Craft

• You can design an interface, prototype it and ship it to production on your own.

• You have strong design taste, and your portfolio shows it: UX, interaction design, information architecture and visual hierarchy. For this role the portfolio is critical.

• You make complex information or a complex data model easy to read. Think maps with metadata or dense dashboards, with non-trivial state. Tools that read and write count for more than infographics or landing pages.

• You have solid computer science and software engineering fundamentals, plus TypeScript, and deep experience building production apps with modern UI frameworks such as React. At least 3 years of delivering production software.

• You understand the problems of engineers, because engineers are CodeSpeak's users.

How you work

• You can say why you made each design choice. "I'm an artist, I see it this way" will not work here.

• You are curious and self-directed, and you care about making programming better. You gather requirements and find projects yourself.

• You actively improve your own workflow with AI-powered coding tools.

• You work hard and can be reached when a task needs it. The team has no weekend-work culture.

• You think fast and explain your reasoning clearly.

• You speak Russian, which is the team's working language. This is a priority, not a hard rule, and a strong candidate without Russian will still be considered.

Nice to have

• A background as a frontend engineer with good taste. Engineers from SwiftUI, desktop, mobile or full-stack are welcome if they visualise complex things very well.

• Experience with complex data visualisation.

• Some sense of what an early-stage startup is like. A big-tech engineer who knows what they are joining can fit, and so can a former co-founder.

A technical degree is not required. Self-taught people and people with degrees in geology or biology have done well in roles like this.

What you get:

• An unsolved problem with real room for your taste. The team is inventing a new way to represent programs, and you decide how it looks.

• A boutique team of strong engineers, built carefully, who have built developer tools at Google, Databricks, Perplexity, JetBrains, Sourcegraph and Elastic.

• An ML team of four, led by Emeli Dral. She co-founded Evidently, an evals startup, and co-authored a very popular Coursera course for data scientists.

• Day-to-day work with Egor, Andrey and the tech lead.

• Backing from Accel and from funds that also invested in Hugging Face and Mistral, with funding for several years.

• A UK contract with a pension and medical insurance.

• Autonomy. You set your own tasks and nobody hovers over you.

• Work in AI, a field most people want to be in.