Build what
matters.

We're looking for someone who thinks about LLM vulnerabilities the way a compiler author thinks about undefined behavior — with rigor, curiosity, and a healthy suspicion of everyone who says "it'll be fine." You'll research AI attack surfaces, write about what you find, and help shape how vsike covers the frontier of AI security.

• Hands-on experience with LLM security: prompt injection, jailbreaks, model extraction, or red-teaming
• Ability to write clearly for technical and non-technical audiences
• Familiarity with the AI landscape: Anthropic, OpenAI, open-weight models, agent frameworks
• Track record of responsible disclosure or published security research

You understand the browser as a platform, not just a target environment for your framework. You can read a CSS cascade error without Googling it. You have opinions about typography and can defend them. You write code that ships fast and lasts long. You'll build and maintain vsike's digital products from the ground up.

• Deep command of HTML, CSS, and JavaScript — frameworks are tools, not identities
• Performance-first mindset: Core Web Vitals, Lighthouse, render-blocking awareness
• Experience building multilingual interfaces (i18n, RTL, CJK typography)
• Accessibility isn't an afterthought for you — it's part of the first draft

Not "make it pretty." We need someone who can think through a product from first principles, prototype with purpose, and ship interfaces that feel inevitable. You obsess over the details that users never consciously notice but would immediately miss. You can work at the intersection of visual design and product logic without losing either.

• Strong portfolio demonstrating product thinking, not just visual craft
• Proficiency in Figma; experience bridging design to production implementation
• Understanding of design systems and when to build vs. borrow
• Comfort working directly with engineers — you can read a pull request

vsike's Ideas section is the edge where AI security meets design thinking. We need a writer who can hold both. You know enough about LLMs to explain them without lying, enough about design to argue for craft, and enough about structure to make a 3,000-word technical post feel effortless to read. Your byline exists because ideas deserve to be clear.

• Portfolio of technical writing that is also genuinely enjoyable to read
• Familiarity with AI/ML concepts: you can explain attention mechanisms without hand-waving
• Research skills: you know how to find primary sources and how to cite them honestly
• Bilingual preferred (EN + ZH or EN + JA) — we publish in three languages

You build things that build things. You've worked with MCP servers, Claude Code, or similar agentic frameworks — not as a user but as someone who extended them. You understand the trust boundary between an AI agent and the systems it touches, and you've thought about what goes wrong when those boundaries are unclear. You'll help vsike build and maintain AI-augmented tooling for content, research, and product.

• Experience building MCP servers, LLM-driven agents, or AI workflow pipelines
• Strong TypeScript or Python; you've shipped code that calls the Anthropic or OpenAI APIs in production
• Security awareness: you think about prompt injection, tool abuse, and least-privilege by default
• Understanding of when AI is the right tool and the intellectual honesty to say when it isn't