Technical Content Lead

The website. Technical blogs, webinars, and product videos. Your own content and following. One person cannot hand-make all of that. So you’ll build the system that can. That isn’t a stretch goal or a second phase - we don’t think the job exists without it. You’ll join the Growth team reporting to a co-founder., * Enough code to be autonomous: Python or TypeScript, APIs, webhooks, a database. Nothing should be blocked waiting for an engineer. We build agentically here - Claude Code, MCP servers, and agent skills are the daily workflow.

• Enough data fluency to be credible: you’re writing for Staff Data Engineers and CDOs. You can hold your own on pipelines, data quality, and warehouses.
• Generalist instinct: you find the constraint and fix it, and you’d rather build the workflow than do the task fifteen more times.
• Independence: you don’t need hand-holding in a distributed, async team across 12+ countries.
• Fluent English.

What you might call yourself. Content engineer, technical content marketer, growth engineer, marketing engineer, GTM engineer, developer advocate, technical writer who got bored, or nothing in particular. We’ve seen this job posted under every one of those. We don’t care what’s on your CV. You need to be able to write and you need to be able to build, and almost nobody can do both., * You’ve grown a technical audience of your own: newsletter, blog, LinkedIn, open source, YouTube.

• Content, growth, or DevRel in data, developer tools, or infrastructure.
• Agentic workflows shipped to real users: MCP servers, Claude skills, tool-using agents.
• You can shoot and cut video yourself.
• Data quality, data management, or data observability industry experience.

• Compensation: $120,000-$130,000 / €110,000-€120,000 per year + equity - the same band wherever you live. We don’t discount for geography.

• Fully remote (US or EU-based), with offices in Brooklyn if you want one.
• All the tokens you need, across all frontier models.
• Real ownership, real impact, no micromanagement.
• Colleagues in 12+ countries.

Our Values We value freedom with responsibility, transparency, and a growth mindset. We avoid politics, have no tolerance for dishonesty, and value open and direct communication without judgment. We proactively bring up and address potential issues before they become critical, and we overcommunicate frequently. Working at Soda, you will have full autonomy to make impactful decisions. ️ Why you will love to join Soda

• Be part of a transformation: data quality is having its moment, and we’re at the front of it.
• Ownership from day one: if you see a bottleneck, you remove it.
• International culture: colleagues in 12+ countries.
• Work on your own terms: fully remote, flexible hours.
• Paid the same wherever you are: one band for the US and Europe, no geographic discount.
• Your name on the work: you build your own audience on company time, and we want you to.
• Direct line to the founders: how Soda talks about the category gets decided with you in the room.

What you might not love

• Rapidly changing priorities: our roadmap can pivot quickly. It’s a fast-paced environment with a LOT of work to be done.
• Flexibility required: you will have to learn new things all the time.
• You will kill a lot of your own output: accurate and lifeless still goes in the bin.
• Judged on output, not architecture: an elegant pipeline nobody reads from is a failure.

Hiring Process * 15-minute phone screen - with a founder

• Short Assignment
• Craft deep dive (90 min)
• Culture fit (1 hour)

We answer every application, and every candidate hears back within 3-5 days of each step. If it’s not a hell yes, it’s a no, and we’ll tell you quickly and kindly.

Data quality is the bottleneck of the AI era. Every Fortune 500 is trying to solve it. We’re the team they call. Soda is the data quality layer for Disney, Ralph Lauren, CBRE, HelloFresh, 2K Games, and Nubank. Our open-source engine, Soda Core, is used by companies like Tesla, Slack, Adyen, Walmart, and JPMC. The category is growing fast because AI doesn’t work on bad data. We’re hiring one person to own content at Soda, end to end. Soda’s mission is to monitor the world’s decisions. We’re building the first Data & AI performance monitoring platform that spans data collection to automated decision-making. Soda helps customers find, understand, and fix data quality issues., Why this is a real problem and not a content calendar. The expertise is already in the building. Founders who have argued about this category for a decade. Customer engineers who spend their week inside broken pipelines at Fortune 500s. Customers who will tell you things nobody has published anywhere. Almost none of it gets written down, and what does get out is hand-made, one piece at a time, by people who have other jobs. Your job is to get it out of their heads and onto the internet, at a volume that is only possible with a machine behind it. Taste is the bar. Anti-slop is the whole point. An engine that ships slop faster has negative value, and most of them do. We’re hiring for the judgment to look at an accurate, competent, lifeless draft and bin it - then work out what the pipeline did wrong and fix it. The engine has to write in distinct voices: a founder sounds nothing like a customer engineer, and the output should sound like them, not like a model. Your own profile and following is part of how we’ll know it works. If it can’t make you sound good, it won’t make anyone else sound good either. What this is not. You’ll write code every day, but this is not a software engineering job. We’re not hiring you to build product, and if you’re a backend, data, or platform engineer looking for your next engineering role, this isn’t it. It’s also not a traditional marketing or DevRel role: no agency to brief and no conference circuit. Requirements

• You can write, and you can tell good writing from bad: send us something you wrote that a technical audience actually read. The most important line in this ad.
• Serious prompt engineering: multi-step LLM pipelines with structured outputs and evals, not chatting with a model. You know why the naive version produces slop.