Slide AI

Job description

Teams can now prompt AI for slide content, but the output is usually brittle, static, and hard to edit. Business and technical teams still need a reliable way to turn structured ideas, data, charts, and visuals into decks that are editable, consistent, reviewable, and ready to present or publish — and prompt-to-static-image tools don't get them there. This role is to build a hosted deck-as-code authoring platform where humans and AI agents create structured deck specs that compile into editable, reviewable, presentable, and shareable slides. The wedge is agent-assisted structured deck generation for repeatable business and technical presentations, using a deck compiler, browser editor, and validation and repair rules to eliminate hand-tuned layout work. ### What You'll Build - A prompt and structured-brief intake flow that produces a deck outline and spec. - A deck compiler that turns the spec into multi-slide output with data-bound chart, math, and AI-generated image primitives. - A browser editor for reviewing and editing generated slides, backed by validation and repair checks for layout, text, connectors, charts, math, and images. - A presenter, publish, and share flow, plus a thin-client CLI or Codex plugin that lets agents create and validate decks through the hosted API. ### What You'll Do - Build the end-to-end hosted authoring app: AI-assisted deck planning, the deck compiler, the browser editor state model, the validation and repair pipeline, and the presenter/publish flow behind a hosted API. - Partner with AI Fund internal teams and 2 to 3 external deck-heavy teams to define pilot scope, success metrics (time-to-first-draft, percent of slides accepted after light edits, layout QA pass rate, publish/share completion), data path, and the MVP cut line. - Design the ladder from prompt-assisted drafting to agent-driven deck creation through the CLI or plugin, keeping every generated deck editable, inspectable, and reviewable rather than a static artifact. - Make decisions about latency, layout validation, repair rules, and rendering reliability so generated decks are trustworthy enough to present and publish. - Work with AI Fund's build team on product wedge, technical risk, and the path from pilot to a broader deck-authoring platform. ### What You'll Need - Strong backend and full-stack engineering: you have shipped production systems involving compilation, rendering, or document processing, with comfort in compilers, parsers, DSLs, or document-as-code, plus React or browser-editor experience. - Demonstrated experience building GenAI products with retrieval, tool use, guardrails, and production feedback loops, including AI-assisted authoring or generation workflows. - Product judgment for human-in-the-loop automation where the right answer is usually to remove busywork rather than replace the human outright. - Ability to iterate with early users and translate ambiguous feedback into working product decisions. - You actively use AI coding assistants and modern GenAI tools as part of your build workflow. - US work authorization. We are unable to sponsor visas for this role. ### Helpful but Not Required - Experience with presentation tools, Figma or design tooling, or visual layout engines. - Background in data visualization, chart rendering, or image generation pipelines. - Experience with CLI or plugin developer tools, marketplace or plugin packaging, collaborative document editing, or publishing workflows. - Founder or founding engineer experience shipping a 0-to-1 product. ### Who This Is For - A strong AI engineer who wants to turn AI slide generation from prompt-to-static-output into an editable structured authoring workflow. - Someone who sees the hard part as making generated decks reliable, editable, validated, and useful in real presentation workflows, not just making prettier templates. ### What to Know Upfront This is a 12-week, full-time, on-site residency in Mountain View, California. Not every residency becomes a company. The goal is to pressure-test the idea quickly and honestly with real users and customers. You will be building an AI Fund idea, not bringing your own startup idea into the program. The process typically includes a Builder Event or equivalent working conversation, then a 48-hour Build Challenge, then panel review with AI Fund build leadership. The compensation is intentionally modest during the residency because the upside, if the idea works, is a founder-level role.