First, which Elise. EliseAI is the New York company formerly called MeetElise (renamed in 2022), founded in 2017 by Minna Song and Tony Stoyanov. It sells conversational AI that runs front- and back-office work for landlords and, since 2023, health systems. Its agents answer calls, texts, emails, and chats around the clock: scheduling apartment tours, handling maintenance requests and lease renewals, collecting rent, and booking patient appointments. If the “Elise” you are looking at is a wellness app or a generic chatbot, this is the wrong one; the company you want is at eliseai.com, built around housing and healthcare operations.
That product shape drives the loop. An engineer here builds agents that act on a customer’s behalf inside their real systems, so the work is as much about reliability and judgment under messy data as about model calls. The closest peer on this site is the Assort Health interview guide, which covers a healthcare voice-agent company with the same “the AI talks to real people and cannot be wrong in dangerous ways” constraint; the Decagon interview guide covers the same rhythm on the customer-support side, and the broader AI-native company interview guides hub collects the rest. Where EliseAI differs is the vertical: leasing offices and clinics, where a wrong answer costs a signed lease or a missed appointment.
What EliseAI is, and why the funding tempo sets the bar
In September 2026 EliseAI raised $350 million at a $4 billion valuation, co-led by Andreessen Horowitz and Bessemer Venture Partners, with Ontario Teachers’ Pension Plan, Sapphire Ventures, and Navitas Capital also in, per the funding coverage. That nearly doubles the $2.2 billion it carried after its $250 million Series E in August 2025, 13 months earlier. Bessemer’s Sameer Dholakia joined the board with the round. The company says it crossed $200 million in annual recurring revenue in June 2026, that its software powers roughly one in six US apartments, and that it serves about 70% of the 50 largest US rental operators. Those are EliseAI’s own figures, but the funding and investor names are confirmed by Fierce Healthcare and mobiHealthNews.
Why this matters at the interview table: the money funds engineering and a second hub in San Francisco alongside the New York headquarters. A company doubling its valuation hires fast, but it guards who it lets near production: its agents run live inside landlords’ and health systems’ operations, and a paying customer churns if the automation misbehaves. The loop is built to find people who can ship against that.
The interview loop, as best anyone can infer it
No official loop is published, and the candidate reports do not perfectly agree, which usually means the exact sequence shifts by role, level, and month. Read the stages below as a composite. One Software Engineer candidate outlined three stages they understood the loop to hold: a 30-minute intro covering technical and behavioral questions, a roughly two-hour live technical assessment, and a system design round that doubled as a culture-and-team-fit conversation with leadership. A separate candidate reported being cut at a first round that was a LeetCode-style coding problem, so for at least some pipelines the first technical gate is a real coding filter, not an intro chat. A senior candidate in New York reported four touchpoints: a behavioral screen with an engineer, a two-hour guided take-home done live over Zoom, an in-person system design round with a VP of engineering, and a final in-person behavioral with the CTO. The two-hour live assessment and the senior candidate’s guided Zoom take-home are almost certainly one stage, so don’t count it twice. A third-party recruiter guide (zerogtalent) says the loop runs four to five rounds over two to four weeks and that the take-home carries heavy weight; treat that as secondary.
The one logistics detail worth settling early is location. EliseAI’s engineering is centered in New York and now San Francisco, and the senior candidate’s design and CTO rounds were both in person, so plan for one of those two cities. Ask on the first recruiter call whether your opening is in-office, hybrid, or remote. On sponsorship: public H-1B records show EliseAI has filed software-engineer visas (about a dozen in 2025), so it is not off the table, but confirm your case early.
EliseAI engineer interview loop at a glance
| Stage (composite from a few candidate reports, one third-party recruiter guide, and job postings, Oct 2026 — not an EliseAI-published loop) | What the reports suggest | How to prep |
|---|---|---|
| Recruiter / behavioral screen (~30 min) | Motivation plus real technical and behavioral questions | A specific reason you want AI for housing or healthcare operations, not generic “I like agents” |
| Coding round, then a ~2 hr live assessment with an engineer | One candidate was cut at an early LeetCode-style problem; a senior candidate’s two-hour guided take-home over Zoom maps to the longer assessment | Practical coding under time pressure; narrate your approach and tradeoffs as you go |
| System design round | Often paired with leadership and a culture/team-fit conversation | Design a reliable agent that acts inside a customer’s systems; lead with failure modes |
| Final behavioral | One senior candidate’s final was in person with the CTO | One project you can defend end to end, told with business impact first |
The coding bar: a real filter, not a formality
At least one candidate was cut at a LeetCode-style first round, so do not treat the early coding as a warm-up chat. Drill the coding patterns catalog until hash maps, sets, two pointers, and the common graph and heap moves are reflexive, and keep your Big-O analysis automatic, because the interviewer is watching whether you reach for the structure with the right worst-case behavior under time pressure. The longer two-hour session is practical rather than puzzle-driven, so it rewards clean, runnable code and clear narration. If any part of it is a take-home, ship it like production: it compiles, it runs, it has a test or two, and a short note explains the tradeoffs. For a company whose agents go live inside paying customers’ systems, legible beats clever.
System design for agents that act on a customer’s behalf
This is where the product shape earns its keep. A realistic prompt, built from what EliseAI does, might be: design the system that lets an agent schedule an apartment tour, or book a patient appointment, by phone and text, end to end, and tell me how you know it did the right thing. A strong answer treats the model as one component in a larger system design, not the whole thing. Start by separating the structured lookups from the open-ended language: availability, unit inventory, and appointment slots are structured data behind a typed tool call or a SQL query, not something you ask a model to recall, and database indexing on the lookup keys is what keeps those reads fast while availability changes all day. Reserve retrieval for the unstructured material: lease terms, policy documents, clinic instructions.
Then spend real time on what makes an acting agent different from a chatbot. The agent writes to someone else’s system of record, so talk about idempotent writes, retries, and confirmation before any booking is committed, because a double-booked tour or a duplicated appointment is a visible failure to a paying operator. Talk about how you measure correctness before the agent touches a real calendar, because “it looked good in the demo” is how you lose a customer. In healthcare especially, name where the agent must refuse and hand off to a human, and how you keep an auditable record of what it did. Candidates underweight that human-in-the-loop and auditability story; bring it up first, because it is close to the whole safety case for the product.
What they screen for beyond code
The behavioral rounds carry real weight, and one of them may be with the CTO. EliseAI sells into landlords and health systems, regulated and reputation-sensitive customers, so the signal they want is judgment: can you reason about what happens when the agent is wrong, can you scope a vague operational pain into something shippable, and can you explain a technical tradeoff to someone who runs leasing offices rather than reads code. Prepare one project you can narrate end to end, with a real decision and a real tradeoff in it, and keep it tight with the STAR method, telling it so the impact leads and the implementation follows. For new-grad and early-career roles the same holds in miniature: show that you understand why reliability matters here, beyond passing a coding screen.
Comp, with the usual caveats
EliseAI’s postings carry real base bands, and they vary widely by level. A 2025-2026 new-grad Software Engineer posting in New York, seen in July 2026, listed a salary of about $100,000 to $180,000, per the Built In listing; that posting may have since closed, so check EliseAI’s careers page for a current opening before relying on it. Recent senior New York postings landed at roughly $230,000 to $300,000 base, a CareerBuilder listing at $240,000 to $300,000 and a Peerlist full-stack listing at $230,000 to $270,000, plus equity and benefits. Self-reported Levels.fyi data puts New York senior total compensation near $428,000 as of an August 2026 snapshot (about $260,000 base and $168,000 in stock), but that sample is thin, so anchor on the posted base bands and treat the Levels figure as a loose ceiling.
The equity is the live variable. Unlike a stealthier startup, you have a public reference point: the September 2026 round priced the company at $4 billion. Grants typically vest over four years with a one-year cliff. Get the strike price, the share count, and the latest preferred price, then model the grant with the total comp calculator and treat it, not the base, as the center of any salary negotiation. The known valuation cuts both ways: the paper value is easy to estimate, but the cheap-entry upside of an early round is gone, so weigh the equity against a higher cash offer elsewhere.
Go in able to build something real with LLMs that you can defend end to end, sharp on the practical coding, and ready to say what your agent does when it is wrong. That last answer separates people who have shipped automation into a real business from those who only demoed it.
Practice the behavioral round:
