Own the two surfaces our customers live in: a real-time coaching extension that renders inside live sales calls, and the manager dashboard where revenue leaders judge their teams — and us. You'll be the front-end owner on a seven-person engineering team, reporting to the engineering lead, at a company one month into go-to-market with live pilots and paying clients.
The bar is the one you know from the best Israeli B2B products: an interface a paying customer trusts on first contact.
Our front-end is the product's moment of truth, and right now it doesn't always meet our bar. We know this precisely, because we track it: coaching cards that appear a beat too late or vanish before the rep can read them, dashboard numbers that don't reconcile across views, filters that look applied but aren't, state that resets on refresh. In a B2B product sold to revenue leaders, every one of those is a critical bug — a manager who catches one wrong number stops trusting all of them. We're not hiring someone to add features on top of this. We're hiring the person who makes "the UI is simply correct" a property of the system.
There's also a hard environment problem most front-end roles never face: our main interface renders inside other people's products. The extension lives in Google Meet and Zoom — pages we don't control, that change without notice, under Manifest V3 constraints, with versions in the field we can't force-update. Shipping reliable UI there is real engineering, not pixel-pushing.
You'll work in an engineering team of seven. Priorities shift when a pilot lands. You'll watch recordings of customers using your UI and feel the misses personally. If you need a design system handed down and a QA department behind you, this is the wrong role. If your instinct on seeing a flickering card is to fix the state model, not the CSS — read on.
The live coaching surface. The extension UI injected into Meet and Zoom: signal cards, captions, the sidebar. Timing is the product here — a card must appear at the right moment, stay readable exactly long enough, and never stack seven deep. You'll own the rendering pipeline from WebSocket message to pixel, including reconnection states and what the rep sees when things degrade.
The manager dashboard. Where sales leaders see team skill gaps, call scores, and roleplay progress — and where they decide whether to renew. Your standard: every number matches every other number, every filter does what it claims, every state survives a refresh. Analytics UI that managers trust enough to bring to their own leadership.
The real-time data layer. WebSocket-driven UI state, optimistic updates that reconcile correctly, pagination and caching that don't lie. The client side of a system where data is always in flight.
Async job UX. Call analysis and roleplay generation run as long jobs. Today users can't always see status, can't cancel, and aren't told about failures. You'll design and build the missing layer: progress, failure surfacing, recovery — so the product never leaves a customer staring at a spinner.
Full-stack when it's faster. The backend is Python/FastAPI. When a feature needs an endpoint shaped differently, you change the endpoint — you don't file a ticket and wait.
The quality bar. Component conventions, visual consistency, regression coverage for the flows customers hit daily. You'll work directly with design and push back when a mock doesn't survive contact with real data.
Customers never see the backend. They judge Yolk — and decide whether AI coaching is real or hype — by the card that appears mid-call and the dashboard their VP opens on Monday. In a seven-person team, front-end quality isn't a department; it's one person's standard applied everywhere. This role decides whether the product feels inevitable or ignorable.
A founding-team seat reporting to the engineering lead — an experienced engineer, not a layer of management — with the CEO a desk away. Competitive salary plus meaningful equity. Performance-based comp and fast advancement. Bi-annual team retreats abroad. Customers who are live and vocal — so your work shows up in revenue, not just in sprint reviews.
A sales rep tracks nine things at once on a live call. Human working memory holds four. Every AI sales tool tries to add more dashboards on top. We subtract — Yolk's AI coach works inside the live call, surfacing the one right thing to say at the moment it matters, then turns what broke down on the call into targeted AI roleplay practice afterward.
The facts: launched end of May 2026. Hundreds of salespeople coached in the first weeks, across several live pilots and paying clients. Thousands of real sales calls already analyzed. Backed by investors in Anthropic and Groq. SOC 2 compliant, 4 patents in core AI methods.
Apply in a few minutes. We read every application and reply to all of them.
Other tools grade the call after it's lost. Yolk is in the room while it counts.