Developer Marketing Intern

About TraceRoot.AI

Open source self-improving layer for AI agents

Tech

TraceRoot is building the open-source self-improving layer for AI agents — observability that detects failures in production, root-causes them against your source code and GitHub history, opens verified fix PRs, and evals every fix — so your agent gets better with every release. Our stack: - Python (FastAPI, Celery) and TypeScript (Next.js 15, React 19) across our platform and agent layers - ClickHouse for time-series trace storage; PostgreSQL for relational metadata; Redis for queuing - OpenTelemetry-compatible Python SDK with native support for OpenAI, Anthropic, Gemini, and 10+ other LLM providers - Agentic root cause analysis engine with GitHub integration for bug tracking and code version history - AWS (EKS Fargate) + Helm/Terraform for deployment; fully self-hostable via Docker Compose

The role

You'll own how TraceRoot talks to developers. Everything we publish — the posts, the launches, the docs that read like arguments, the demos — is yours to write and ship. Our growth today comes from developers who found the repo, installed it, and told someone. Your job is to make far more of them find it.This is a writing job for someone who likes software. You don't need to be an engineer, but you do need to be the kind of person who installs the tool, breaks it on purpose, and finds the story in what happened. The best thing we can publish is a developer explaining something true and specific about debugging agents — that's the bar, and you'll hit it weekly.We move fast, debate hard, and execute even harder. You'll work in person with me at least 3 days a week. You won't own everything in one summer — you'll pick a channel with me and own that number end-to-end. There's a path into our full-time Founding GTM role; be aware that role adds calls and pipeline to the writing, so converting means growing into the commercial half too.## What You’ll Do* **Write the things developers actually read** — technical deep dives, postmortems, and docs-as-marketing about what breaks in production agents and how to see it. Not thought leadership; specific, useful writing with real traces and real code in it.* **Run the launches** — Hacker News, Product Hunt, Reddit, and X. You'll write the post, prepare the assets, be in the comments answering people all day, and do the follow-through that turns a spike into retained users.* **Own SEO and GEO** — comparison pages and problem-first pages that rank, and getting TraceRoot cited by ChatGPT, Claude, and Perplexity when someone asks how to debug an agent. This is a compounding channel and it will be substantially yours.* **Show the product, don't describe it** — short demo clips, screenshots, and walkthroughs that make the thing obvious in ten seconds; plus dev-influencer collabs, hackathons, and a Discord where people actually get answers.* **Know what worked** — track which post, page, or launch actually drove installs and traces, and say so honestly, including when the answer is "none of them." We'll set you up in PostHog; you don't need to build it.## Requirements* **You write, and people read it:** a blog, a newsletter, a thread that went somewhere, docs, a README people thanked you for. Any of it counts. Send it.* **Technically curious enough to be credible:** you can install TraceRoot yourself with Claude Code, read a stack trace, and explain what happened without a hand-holder. You do not need to write production code for this role.* **You have taste in developer content:** you know why a post lands on Hacker News and why marketing copy gets torn apart there, because you read those threads for fun.* **Fast and prolific:** you'd rather ship six pieces and learn which two worked than polish one for a month.* **Strong sense of ownership and bias for action:** you own outcomes, not tasks, and you make things happen without waiting for a roadmap or a content calendar.## Preferred Experience* **You've done this unpaid already:** grown an open-source project, run a technical blog or newsletter, or written docs for something people use.* **In the community:** you're around people building agents, and you know who the voices are.* **LLMs as a tool, not a topic:** you use model APIs to research, draft, and repurpose — and you know exactly where that stops working and you have to write it yourself.## To ApplySend two things:* **Two sentences on the first experiment you'd run** to get more agent-building teams onto TraceRoot.* **A link to something you built** — a repo, a script, a scraper, a landing page, a Discord bot, an outreach system. It does not need to be a growth project and it does not need to be impressive; we just want to see you ship.

Skills

• Python
• SQL
• Data Analytics
• Growth design
• Marketing design