Magic (makers of Loyalist) · New York, NY · In-office 4 days/week · $150k-$200k + equity
I’m Niky, a founding engineer at Magic. You’d be sitting next to me if you took this job, so I wanted to tell you about it from my perspective.
A bit about how I got here: I was working at a clinical data AI startup, and on the side I was reading every hospitality book I could find. I had a half-built product design spiraling around in my head. Then I discovered Magic: Maggie was already building it. I felt a very specific kind of disappointment followed by a very specific kind of exhilaration. I cold-emailed Maggie the next day and joined six weeks later.
Loyalist is the operating system that powers how our favorite restaurants remember you.
We ingest live data from ~30 third-party systems during service (reservations, POS, reviews, private events, marketing, e-commerce). We turn that data into guest profiles and automated actions: a server hands a VIP a comped glass of wine because the system flagged that she’s been posting five-star reviews over the last year. A marketing manager sends a happy-birthday offer to 2,000 guests at once, each message referencing the specific dish that guest ordered last time. A sommelier gets a heads-up that a two-top walking in tonight are repeat 2018 Barolo drinkers.
800+ restaurants run on this today, from neighborhood favorites to Carbone, Momofuku, Le Bernardin, and Daniel Boulud’s group. Some of the biggest have 5-10 million guests each and hundreds of millions of visits and orders going back 15 years. We process more than 10 million guest interactions every year.
The product looks like a CRM, but under the hood, it’s closer to event-driven infrastructure stitching together 30+ third-party systems in real time.
A few of the problems I find most interesting:
Customer journeys built off operator intuition. Operators come to us with messy, specific requests: a flow that goes out to lapsed regulars on the day of the week they used to come in, a birthday email that references the wine they ordered last time, a follow-up triggered by a recent five-star review. We turn those into running journeys, with branching logic and tags that follow a guest through every interaction. Right now we mostly build the journeys with operators; we’re moving toward letting them build their own.
Tools that run during service. A regular texts our SMS line and the GM responds before they sit down. A server pulls up tonight’s pre-shift report and sees which VIPs are coming in, what they drank last time, and which ones haven’t been back in 90 days. This stuff has to work flawlessly when there are 200 covers in service and the kitchen is on fire.
Schema flexibility across 800 idiosyncratic customers. Before I joined, I thought Loyalist was one product sold to 800 restaurants. It’s closer to 800 subtly different configurations of the same product, because every group has its own way of capturing preferences, tagging VIPs, or defining who counts as a “regular.” Our schemas and data models have to be strict enough to hold 40 million guest profiles reliably and loose enough to honor each group’s definitions. (Hotels are next, where the word “guest” means something different yet again.)
You’d be a fullstack engineer shipping end-to-end features for restaurants. Engineers own the full arc here: talking to the customer, scoping the work, and shipping it. We’re also hiring our first designer, who’ll partner with engineering on craft and polish. But the customer relationships and feature decisions stay with you.
That sounds intimidating, and it kind of is. But it also means you get to be the one on the customer call understanding the problem, and the one at your laptop solving it. That’s the part of the job I love most.
A feature I shipped recently that I’m still pretty proud of is Journeys, our system for designing multi-step guest experiences across multiple venues and integrations. Operators define what should happen across a guest’s lifecycle (welcome, post-visit, lapsed-regular, anniversary), and Loyalist orchestrates it across reservations, POS, marketing, and reviews—in sequence and in parallel, with every step idempotent and recoverable.
The biggest strategic direction right now is moving onboarding from engineer-led to operator-led. We’ve historically configured each customer’s Loyalist instance by hand (which is how I learned the industry). We want operators to drive most of that themselves. That’s a lot of UI work with real consequences, and much of it is what you’d be working on.
I’ve been on both sides of the interview table here, so I’ll be direct.
You’ll probably do well if: