Website:
100xlt.ai
Job details:
Company Description100XLT.ai empowers local publishers by providing seamless solutions to capture a greater share of the $47 billion in local ad spend that often goes to major platforms like Google and Meta. We offer a hassle-free way for publishers to create advertising campaigns that support local businesses while boosting their own revenue. Our platform enables local businesses to have a stronger community presence through publishers' networks, creating value for both parties. With growing adoption among major publishers and regional media operators, 100XLT.ai is revolutionizing local advertising. At 100XLT, our success is inherently tied to the success of our partners.
About the Role
We build AI-powered content products, and our engineering team ships through orchestrated AI agents rather than hand-typed code. This role owns the surface our users actually touch — the CRM where publishers build campaigns, edit AI-generated content, review ads, and see their numbers.
This is not a "convert the Figma" job. You'll own a shared design system consumed by multiple apps, a collaborative rich-text editor built on TipTap and Yjs, data grids that hold real volume, and UI that streams and renders LLM output safely. These are the hard parts of frontend, and they're yours.
The role is less about how fast you type and more about how well you think, decompose, and verify. You burn tokens to deliver impact, and you know exactly when it's worth it.
The one-line test: Can you take a vague feature request, spec it, orchestrate agents to build the UI, verify it's actually correct — types, states, edge cases, accessibility — and ship it before a traditional team finishes grooming the ticket? If yes, apply.
This is an on-site position — we move fast, pair tightly, and think best in the same room.
A note on experience: the years-of-experience bar below is a guideline, not a gate. If you have genuine AI-first build experience — you've built agents, harnesses, or automations that made you or your team dramatically faster — we want to talk, even if you're earlier in your career. Demonstrated agentic output beats a long résumé.
What you'll actually do
- Own frontend features end-to-end — from a vague request to production — across a TypeScript monorepo with a React SPA, a shared component library, and a shared theme package.
- Grow and defend the design system. When a page needs something new, you decide whether it belongs in the shared library or the page, and you make that call well.
- Build the hard surfaces: a collaborative rich-text editor (TipTap/ProseMirror + Yjs), data grids, multi-step campaign flows, and UI that streams LLM output.
- Decompose ambiguous UI problems into specs an agent can execute, then orchestrate, review, and tighten the loop.
- Build agents, workflows, and automations that push the whole team's frontend velocity toward 100X.
- Consume a generated API contract. Our types come from the backend's OpenAPI spec via codegen — you work with that boundary and push back on the backend when the contract is wrong.
- Hold a high bar — correctness, accessibility, performance, and maintainability — while moving fast. Speed without judgment isn't what we're after.
Must-haves
Behavioral
- Builds agents to multiply the team's output. You don't just use coding agents — you build them: custom agents, workflows, harnesses, and automations that take a team from 1X to 100X. When a task is repetitive or parallelizable, your instinct is to make an agent do it.
- Output measured in shipped impact, not effort. You ship features, not lines.
- Lives in agentic tooling. Daily, fluent use of Claude Code / Cursor / equivalent — with real opinions on getting more out of them.
- Verifies relentlessly. You read what the agent produced, run it in the browser, click through the states it didn't think about, and own the result. No blind merges.
- Decomposes well. Turns fuzzy asks into clear, executable specs — this is the core skill.
- Spends tokens like an investment. Comfortable fanning out agents aggressively, and able to judge when it pays off.
- High agency. Unblocks yourself, fills gaps, and pushes things to done without hand-holding.
Technical
- Senior-level command of TypeScript and React (typically 4+ years building production frontends — but strong AI-first builders with less time are encouraged to apply). Strict TypeScript is not optional here: any is a build error, and you should reach for unknown plus type guards without being asked.
- Deep with a modern client-side data layer — TanStack Query or equivalent. Query keys, cache invalidation, mutations, optimistic updates, and the failure modes that come with them.
- Real experience with a typed router — TanStack Router, or enough React Router / equivalent depth to pick it up in days. Typed params, route-level loading, code splitting.
- Tailwind and headless/primitive component libraries (Radix, react-aria, shadcn-style patterns). You can build an accessible component from primitives, not just style a div.
- Design-system instincts. You've built or maintained a shared component library and know why a page should extend it rather than fork it.
- Forms at real complexity — react-hook-form + Zod or equivalent. Validation, dirty state, unsaved-changes guards.
- Testing that survives speed — Vitest / Jest with React Testing Library, and the judgment to know what's worth testing in a UI.
- Modern frontend toolchain — Vite, a package manager with workspaces (we use pnpm), and comfort in a monorepo with shared packages and build orchestration.
- Has shipped at least one LLM-powered feature end-to-end, including the frontend reality of it: streaming, partial states, retries, and rendering model output without opening an XSS hole.
- Debugging and security instincts that hold up under speed — you know why user-generated and model-generated HTML gets sanitized before it renders.
Good-to-haves
- Rich-text editing — TipTap, ProseMirror, Lexical, Slate. Custom extensions, not just the starter kit.
- Real-time / collaborative editing (CRDTs) — Yjs, awareness/presence, conflict handling.
- Data grids at scale — AG Grid, TanStack Table, virtualization, server-side pagination.
- Worked in a monorepo with type-safe codegen between backend and frontend (OpenAPI → TypeScript).
- State machines for complex flows — XState or equivalent.
- i18n in a real product — namespacing, extraction, translation workflow.
- Charting and dataviz — Recharts, D3, or similar.
- Frontend performance work — Core Web Vitals, bundle budgets, code splitting on a large SPA.
- Auth on the client — Cognito/Amplify, OAuth, token refresh, protected route trees.
- Built internal tooling that made a team measurably faster.
- Contributions to open-source agent tooling or MCP integrations.
How we work
- Ship-daily culture, tight feedback loops, and AI-orchestrated review. We give you the most capable models, generous token budgets, and the autonomy to use them. Bring the judgment.
Click on Apply to know more.