Fly.io Interview Guide (2026): No Resumes, No Whiteboards, Just Work-Sample Challenges

Fly.io

fly.io ↗

Updated · techinterview.org

Fly.io is a developer-focused public cloud that runs your apps in Firecracker microVMs on hardware in dozens of regions close to users, with a stack built largely in Elixir, Go, and Rust. It was founded around 2016-2017 by Kurt Mackey (who previously co-founded Compose, the MongoDB hosting company) along with Michael Dwan and others, and has raised roughly $110M total, including a $70M Series C in 2023 that reports put at a valuation just under half a billion dollars, with backers including Andreessen Horowitz and Intel Capital. The company is small (around 60 people as of 2026, after a round of layoffs in 2024) and famously opinionated, nowhere more so than in hiring: their blog and public docs argue that resumes and interviews are bad predictors, so the entire process is built around graded work samples. If you are used to grinding leetcode for infrastructure roles, this loop will feel like a different sport.

Process

The process is documented on Fly.io’s own hiring docs, so there is less guesswork here than with most companies. It goes like this: you email the address at the bottom of a job posting (no ATS portal), someone replies to lay out the challenges and answer questions, and then you receive a series of at-home challenges, usually two or three, delivered one at a time. Their docs say the challenges are extracted from work the team has actually done, simplified to a few hours each, and graded against an objective rubric. There are no timers and no screen-sharing; their docs explicitly say “take a day, take a week, take a month,” though they note staffing needs can change if you stretch it out. Your resume is optional and, per their docs, not part of the decision. If your challenge submissions score well, the final step is a 60-90 minute structured conversation held entirely in Slack, in text, with no video or audio, followed by an offer. Total cycle length is largely under your control: candidates who move quickly report a few weeks end to end. One caveat from Hacker News threads: some candidates say challenges advertised as “a few hours” took closer to ten, so budget more time than the pitch suggests. A few Glassdoor reviews also describe rejections as abrupt, since rubric grading means you can be cut after one challenge with limited feedback.

What they actually ask

  • Platform and backend challenges in Go or Rust that mirror their proxy and orchestration work: candidates describe tasks like building a piece of a TCP/TLS proxy or a small service that manages VM-like workloads. Fly.io has published some challenge repos under a public “fly-hiring” GitHub org, so you can often preview the shape of the work before applying.
  • Distributed systems challenges for senior and staff roles: Fly.io publicly released “Gossip Glomers,” a Maelstrom-based series (unique ID generation, gossip broadcast, replicated logs, and similar) that they built specifically because they lacked a good staff-engineer signal. Working through it is the closest thing to a leaked exam this company has.
  • Full-stack and Elixir challenges: feature work against a realistic Phoenix/LiveView app rather than an abstract puzzle, matching their blog posts recruiting full-stack developers.
  • SRE work samples: their SRE hiring project has circulated publicly on GitHub and involves operating and debugging a realistic deployment rather than answering trivia about load balancers.
  • Support engineering samples: a public “support-engineer-work-sample” repo has candidates write responses to realistic customer tickets, which tells you how much this company values written communication at every level.
  • The Slack structured interview: expect written questions about the decisions you made in your challenges, how you debug production problems, and how you would talk to users. It is graded conversation, not small talk, and your writing quality carries real weight.

Levels and comp (2026)

Fly.io publishes an unusual comp policy: everyone at a given level gets the same salary and options, regardless of role, geography, or prior experience, and they do not negotiate, on pay-equity grounds. They also state they pay the same US-based schedule worldwide, which is a notably good deal for candidates outside the US. Concrete numbers are thin because the company is small: a platform engineer posting listed roughly $190K to $225K base depending on level, levels.fyi shows a handful of software engineer reports with total comp clustering around $200K to $230K, and Glassdoor has fewer than ten salary entries. Treat all of these as directional rather than precise. Since negotiation is off the table, put your energy into the level conversation instead: ask how they decided your level, what the next level requires, how many options you are getting, the strike price, and the most recent preferred price from the Series C, so you can value the equity yourself. Equity at a sub-$500M-valuation infrastructure company is a real lottery ticket, in both directions.

Prep priorities

Skip the algorithm grind; there are no binary trees to invert here, and their docs say so almost verbatim. Instead, practice building small networked systems in Go or Rust: a toy TCP proxy, a rate limiter, a job runner that supervises processes. If you are applying for senior roles, work through Gossip Glomers before you apply, since it doubles as their published staff-level bar, and brush up on system design fundamentals like consensus, replication, and failure modes, because the rubrics reward reasoning about what breaks. For full-stack roles, get comfortable shipping a feature in Phoenix/LiveView end to end. Then practice writing: the final round is a text-only Slack interview, so your ability to explain a technical decision in three clear paragraphs matters more than at almost any other company. If your fundamentals are rusty, a pass through common coding patterns helps you write cleaner challenge code, but treat it as a warm-up, not the main event. Finally, prepare a few stories about ownership and shipping under ambiguity for the structured conversation; the same material you would build for a behavioral round translates directly into written form.

Frequently Asked Questions

Does Fly.io really ignore resumes? Their docs say resumes are welcome but play no role in decisions; grading is done on challenge submissions against a rubric. Candidates’ reports generally back this up, though the initial email you send still creates a first impression, so write it well.

How long do the challenges actually take? Fly.io says each is scoped to a few hours, and delivery time is not graded. Some candidates on Hacker News report spending closer to ten hours on a single challenge, so assume the realistic total across two or three challenges is a couple of working days of effort.

Can I negotiate the offer? No. Comp is standardized per level and they state they do not negotiate, to avoid rewarding negotiation skill over job skill. The one thing you can influence is level placement, so ask what evidence would support a higher level before you finish the challenges.

Is the company fully remote? Yes. Fly.io is remote-first with no central office, hires globally, and pays the same schedule everywhere. The whole process, including the final interview, happens over email and Slack, so you may get hired without ever being on a video call.

newsletter

What's actually being asked right now

Interview patterns & comp trends, straight to your inbox.

No spam. Unsubscribe anytime.

newsletter

What's actually being asked right now

Interview patterns & comp trends, straight to your inbox.

No spam. Unsubscribe anytime.

1972 Soviet postage stamp commemorating the Mars 2 probe

worth a read

Mars For The Rest of Us — a weekly-or-more deep dive on the technical side of Mars exploration: rocket propulsion, microbiology, mission architecture, and everything in between. Written by Maciej Ceglowski.

Read it on Substack
Scroll to Top