Head of virality

Work that outlasts a career. We're solving one of the most ambitious problems in AI: building an AI to understand and be fully personalized to humans and organizations.

We're growing like crazy and it's one in a lifetime opportunity to join the next hockey-stick-curve startup before the singularity.

Join a world-class team of engineers building technology that could reshape AI.

We ship three product demos every week. Each one shows a real workflow, in the real product, with real proof on screen. Building that cadence is the job.

screenpipe records what you see, say, and hear locally, so the AI you already use can work from what actually happened. It is unusually demo-able: you can show software observing work and turning it into a dashboard, SOP, automation, or invoice someone can defend.

Most people have never seen anything like it. Almost nobody knows we exist yet. That gap is the opportunity.

What you'll own

• Demos every week, end to end: choose the workflow, write the story, record it, edit it, publish it, and distribute it.
• Help the founder build a company cult
• Founder-led channels across X, LinkedIn, YouTube, TikTok, and Instagram.
• Turning real customer workflows into content instead of inventing hypothetical use cases.
• Interviewing users and finding the moments that make someone stop scrolling.
• Measuring qualified inbound, activation, reposts, and pipeline - not vanity impressions.
• Killing formats that do not work and scaling the ones that do.

What good looks like after 90 days

• People who do not know us repost our demos.
• We have repeatable formats that work without a production agency.
• Product launches have a reliable content and distribution system.
• Sales receives qualified inbound that can become revenue.
• The best-performing content teaches the product team what to build next.

You

• You have made product demos that spread. Send the links - that is the first screen.
• You shoot, edit, write, and publish yourself.
• You understand products deeply enough to show the real mechanism rather than hiding it behind cinematic editing.
• You have excellent taste but ship before things feel perfect.
• No agency dependency, committee-driven content, or AI slop.
• You are in San Francisco and want to work in the room with the product.

screenpipe is YC S26, a team of two, and early enough that your work will define what the company sounds like.

• Health and recovery: Personal health coach or trainer, gym membership, sauna and cold plunge access, and unlimited massages.
• Health insurance: Top-tier health insurance.
• Flexibility: San Francisco is our home base, with flexibility to work remotely from time to time, including trips abroad such as Thailand.
• Learning: Unlimited books and audiobooks, on us.
• AI tools: $20,000 per month in AI tokens to build, experiment, and do your best work.
• Personal growth: Career and life coaching.
• Team travel: Team trips and work retreats, including places like Hawaii.

We look for curiosity, independent thinking, direct and thoughtful feedback, and ownership of real user outcomes. We want people who investigate problems, build useful things, and check whether their work actually helped.

When you apply, include short answers to these questions alongside examples of your work. Bullets and links are welcome. Estimates are fine; explain what the numbers represent.

1. Reading and curiosity: Roughly how many books did you read or listen to in the past year? Which two or three are your favorites, and what idea from one changed how you think or work? What are you learning now?
2. AI usage: Roughly how many tokens do you use in a typical week, across which models and tools? If your tools do not expose token counts, share your approximate weekly AI spend or plan and usage pattern instead. Separate personal usage from a team's total, and describe what you produced with it.
3. Your AI setup: Walk us through the setup you actually use: models, coding agents, editors, memory, MCP servers, skills, and automations. Which parts did you configure or build yourself? A sanitized excerpt from your CLAUDE.md, AGENTS.md, or equivalent instructions is welcome.
4. An unusual workflow: Show one distinctive AI workflow you use repeatedly. Explain its inputs, steps, tools, output, and how you check quality. What did it replace, what improved, and where does it still fail? A short demo, diagram, or concrete example works.
5. Ownership and user judgment: Describe an ambiguous problem you took from discovery to a shipped result. What did you learn directly from users, what did you decide not to build, and how did you know the result helped? Be clear about your own contribution.
6. Truth and feedback: Tell us about evidence or feedback that changed a strongly held product or technical opinion. How did you respond, and what changed in your work or collaboration?
7. Initiative: What have you built or improved because you thought it should exist, without someone handing you a detailed task? What would you investigate first at Screenpipe, and why?

We care about how you learn, exercise judgment, and produce useful outcomes. Share public or anonymized examples only; remove credentials, private prompts, and confidential customer or employer information.

We run a desktop app that captures continuously and reliably on people's machines, all day, without getting in the way. That one constraint makes almost everything hard and interesting.

The stack:

A Rust core for high-throughput screen and audio capture, accessibility-tree parsing, and local indexing for fast search. A Tauri desktop app in TypeScript and React, shipping on macOS and Windows. On-device AI. We train and run our own models locally, including PII redaction, so sensitive data never has to leave the machine. Local-first storage and a local API that agents and pipes build on top of. Hard problems we live in: capturing everything without slowing the machine down, running ML on-device across wildly different hardware, turning messy accessibility trees into clean structure, and staying stable enough that people trust us with their entire digital life. We build in the open.

We are training our own privacy model and foundation model to understand human work activity in a modality frontier companies will never risk, which is long term screen recording.

1. Send links to product demos or launch content you personally created.
2. 20-minute founder conversation about your contribution, distribution, and measured outcome.
3. Working session using screenpipe: identify a real workflow and outline how you would turn it into a demo.
4. Final conversation covering cadence, ownership, and the first 30 days.