# Tennr Interview Guide (2026): Take-Home Build and Document-AI Product Sense

Source: https://www.techinterview.org/companies/tennr-interview-guide/
Updated: 2026-09-08 · techinterview.org

Tennr is a New York healthcare-automation company that reads faxed referrals and orders with its own vision-language model (RaeLM) and pushes the patient through intake, eligibility, payer qualification, and scheduling. The product and backend engineering loop is not published; candidate reports for the implementation-engineer track show a 48-hour take-home, a two-problem coding round, a short system design round, and a behavioral call, and Glassdoor puts the average process at about 11 days. Posted bands on the Ashby board run $165K to $215K base plus equity and a stated bonus range; levels.fyi has thin data near $260K total for a senior.

Tennr was founded in 2021 by Stanford classmates Trey Holterman (CEO, previously an engineer at Strava), Diego Baugh (CPO), and Tyler Johnson (CTO, who leads engineering), and went through Y Combinator's Winter 2023 batch. It sells software that ingests referrals and orders arriving by fax, phone, email, and portal, extracts what matters, checks eligibility and payer rules, and drives the follow-up so the patient gets seen. The company raised an $18M Series A led by Andreessen Horowitz in early 2024, a $37M Series B led by Lightspeed in October 2024, and a $101M Series C led by IVP in June 2025 at a reported $605M valuation, with GV, ICONIQ, and Frank Slootman joining. Fortune reported eight-figure revenue at the Series C. Headcount depends on the tracker: YC lists 205, Tracxn recorded 311 in June 2026, and the company signed a 125,000-square-foot lease at 345 Hudson Street in Hudson Square that spring.

## Process

Tennr has not published its engineering loop, and Glassdoor's roughly three dozen reviews skew toward implementation-engineer and go-to-market candidates, so treat what follows as inference. The one detailed engineering report, an Implementation Engineer candidate on LeetCode Discuss in August 2025, describes four stages: a take-home due in about 48 hours (design a workflow that pulls a table from Wikipedia and appends it to a Google Sheet through the APIs), a live [coding](/algorithm-patterns-cheat-sheet/) round with two easy-to-medium problems, a [system design](/category/system-design/) round that mixed a rate-limiter design with short-answer questions, and a [behavioral](/post/3233460379/behavioral-interview-questions-2026-star-method-amazon-leadership-principles-and-winning-answers/) conversation about background and motivation. Glassdoor reports an average of 11 days to decision, a 2.8 out of 5 difficulty score, and a low positive-experience rate, with complaints about thin feedback after the take-home.

The New York product-engineer and backend postings ask for three to five years of experience, JavaScript with TypeScript, experience wiring ML models into production workflows, and a "founder's mindset," and every engineering role is on-site. Expect the same skeleton for those roles, with the take-home carrying more weight and a founder or CTO conversation at the end. Unlike the [Abridge](/companies/abridge-interview-guide/) loop, this is a product-engineering loop where the model is a component you integrate rather than the thing you are hired to invent. It sits mid-pack on the [AI startup difficulty index](/ai-startup-interview-difficulty-index/): the algorithms are easy, but the take-home and the domain questions punish people who have never shipped a workflow product.

| Stage | Who | What it tests |
| --- | --- | --- |
| Recruiter screen (30 min) | Recruiter or hiring manager | Why Tennr, why healthcare operations, comfort with an on-site week in New York |
| Take-home build (about 48 hours, reported) | You, then an engineer reviews | Pulling data from one API and writing it to another cleanly, with error handling, tests, and a readable README |
| Live coding (45 to 60 min) | Peer-level engineer | Two easy-to-medium problems; reported examples are Top K Frequent Elements and a Valid Parenthesis String variant |
| System design (45 to 60 min) | Senior engineer | Rate limiter high-level design, SQL vs NoSQL vs Redis, containers vs VMs; expect document-pipeline variants for product roles |
| Behavioral and founder conversation | Hiring manager, often a founder | Ownership, how you handle ambiguity, how you have learned a domain from scratch |

## What they actually ask

- Design a workflow that pulls a table from a Wikipedia page and appends the rows to a Google Sheet through the APIs. Handle schema drift, rate limits, and reruns without duplicating rows. This is the reported take-home; reviewers care about idempotency and the README more than cleverness.

- Given a stream of events, return the K most frequent elements, then explain the time complexity of your heap versus a bucket-sort approach and when you would pick each.

- A string contains parentheses plus wildcard characters that can stand in for either bracket or nothing. Decide whether it can be made valid, then extend it to brackets of multiple types.

- Sketch a rate limiter that protects a payer eligibility API we call on behalf of hundreds of customers. Where does the counter live, what happens when Redis is unavailable, and how do you stop one large customer starving the rest?

- A faxed referral arrives as a 14-page scanned PDF with a handwritten prescription on page nine and an insurance card photocopied sideways. Walk through how you would turn that into a structured order, where the model sits, and what you do when the extraction confidence is 88 percent.

- Our automation only fires above a confidence threshold that customers pick themselves. How would you build the review queue for everything below the line so reviewers are fast, the model learns from their corrections, and no protected health information leaks into logs or prompts?

- Tell me about a time you owned a feature from customer complaint to shipped fix, and what you would do differently. Then tell me why you want to spend your week on fax intake for sleep clinics and durable medical equipment suppliers.

## Levels and comp (2026)

Tennr posts salary bands on its Ashby job board, the most reliable comp source for this company. The Backend Software Engineer and Product Engineer listings (New York City Office, on-site) state "$165K – $190K • Offers Equity • $33K – $57K Bonus." The Senior Product Engineer and Senior UX Engineer listings state "$190K – $215K • Offers Equity • $38K – $64.5K Bonus." Listings mirrored on the a16z jobs board show Senior Forward Deployed Engineer and Enterprise Solutions Engineer at $160K to $180K. Levels.fyi lists a median software-engineer package near $260K total on very few submissions; the visible senior entry is $220K base plus a $40K bonus with no stock recorded, which more likely means the submitter left equity blank. Glassdoor and Blind have almost no engineer comp data for Tennr as of September 2026, so anchor on the posting and negotiate against the top of the band.

Equity is the part to interrogate. The last priced round was a $605M valuation in June 2025 with about $162M raised in total, so a meaningful preference stack sits ahead of common. Ask for the strike price, the latest 409A, the vesting cliff, the post-termination exercise window, and the fully diluted share count. The stated bonus range is unusual for a company this size; ask whether it is a company-performance target, an individual metric, or a signing bonus, because those are worth very different amounts. A private mark is still a private mark.

## Prep priorities

Learn the referral workflow first, because every technical question at Tennr is a version of it. A specialty provider (a durable medical equipment supplier, a sleep lab, an infusion center, more recently orthopedics, cardiology, and fertility) receives an order by fax; someone must read it, confirm insurance and benefits, check the order against payer medical-necessity criteria, chase the referring doctor for missing documentation, and schedule the patient. Tennr's modules map onto those steps, Tennr Network adds status visibility for both sides of the referral, and the voice agent announced in January 2026 makes the outbound benefits and confirmation calls. Read the Healthcare AI Guy profile of Tennr and one payer's DME coverage policy so you can say what "qualification" means with a concrete example, and know what HIPAA de-identification implies for training data and prompts.

On the technical side, build a small TypeScript pipeline that calls two third-party APIs, retries, deduplicates, and ships with tests before you apply, and reuse its structure for the take-home. Refresh medium-level array, hash map, heap, and string problems; the reported questions are LeetCode staples and the bar is clean code under conversation. For design, practice queue-backed document pipelines, confidence-gated human review loops, and multi-tenant rate limiting, and be ready to discuss the tradeoff Tennr's leadership describes, where 99 percent precision automates a fifth of volume and 97 percent automates half. The [study plan](/study-plan/) covers two weeks of this. Ask the recruiter which track you are on (product, backend, forward deployed, or implementation), whether the take-home is reviewed live, and the bonus mechanics. Ask how much of the stack is fine-tuned open-weight models versus vendor APIs; reporting says Tennr fine-tunes Mistral and Llama variants, and the answer tells you what a product engineer here touches day to day.

## Frequently Asked Questions

### Does Tennr require a take-home assignment for engineers?

Every public candidate report includes one, due in about 48 hours, with an API-to-API pipeline prompt (Wikipedia table into Google Sheets). Product and backend candidates should expect the same shape, possibly with a document-parsing twist. Reviewers complained about slow feedback after submitting, so ask for a decision date before you start.

### Is the Tennr interview hard?

Glassdoor rates the difficulty at about 2.8 out of 5, and the reported coding problems are easy to medium. The harder parts are the take-home, where craft matters, and the design conversation about document pipelines, confidence thresholds, and protected health information. Workflow and integrations engineers do well; pure algorithm grinders stumble on the domain.

### Is Tennr remote or in-office?

All current engineering postings are tagged on-site at the New York City office, and Glassdoor reviews describe four days a week in Hudson Square with lunch provided. Implementation-engineer roles have been listed remote, including outside the United States, but that is a different track with different pay.

### What does Tennr pay software engineers?

Posted Ashby bands are $165K to $190K base for Backend Software Engineer and Product Engineer, and $190K to $215K for Senior Product Engineer and Senior UX Engineer, each with equity and a stated bonus range. Levels.fyi shows a median total near $260K on a handful of entries. Verify the band on the live listing; the numbers have moved up at least once since the Y Combinator profile was written.