Welcome. This repository is how we at Raft evaluate candidates across engineering and mission-oriented roles. Every challenge here is a real, scoped problem that mirrors the kind of work Rafters do day-to-day — shipping focused software that helps mission users make better decisions, faster.
If you were sent a link to a specific folder under challenges/,
jump there and read the README. The rest of this page is context on who
we are, what we build, and why we work the way we do.
Raft is a mission-driven technology company. We build software for the U.S. government and its mission partners — the operators, analysts, and warfighters who make consequential decisions every day and who deserve modern tools to do it with.
We work in the hard spaces: data that lives in dozens of incompatible systems, AI that has to be safe and explainable, applications that have to ship to operational environments with real constraints. We bring commercial-grade engineering practices — small teams, fast iteration, outcome-driven delivery — to problems that historically haven't gotten that treatment.
We move fast and we're honest about what's working and what isn't. We pair engineers tightly with mission users so feedback loops are short. We invest in each other — Rafters pick up new stacks, move between products, and teach what they know. And we make time to have fun doing all of it.
See
assets/images/README.mdfor the image slot list and naming convention.
Raft ships three platforms that work together. Each one solves a specific class of mission problem, and together they form the spine of how we deliver modern software to federal users.
The data foundation. [R]DP handles ingest, governance, cataloging, and interoperability for mission data — bringing together sources that typically live in isolated enclaves and making them queryable, discoverable, and trustworthy. It's the layer everything else we build stands on.
AI, built for the mission. [R]AIMS turns mission data into decisions. It's how we ship models, agents, retrieval systems, and evaluation pipelines into environments that care deeply about safety, auditability, and performance. [R]AIMS sits on top of [R]DP and feeds insights up into applications.
Apps that mission users actually use. [R]AP is the delivery layer — the framework, components, and deployment primitives we use to build focused mission applications quickly, without reinventing the wheel on every engagement. Apps built on [R]AP consume [R]DP data and [R]AIMS intelligence to put the right answer in front of the right user at the right moment.
Challenges are organized by track under challenges/:
| Track | Who it's for |
|---|---|
ai-ml/ |
ML engineers, applied scientists, data scientists |
full-stack/ |
Full-stack engineers |
backend/ |
Backend / platform engineers |
frontend/ |
Frontend engineers |
sales-engineering/ |
Sales engineers, solutions engineers |
See challenges/README.md for the full index
of active challenges.
These challenges are scoped to respect your time. If a challenge says "no more than 4 hours," we mean it — we're evaluating your judgment, not your endurance. Make reasonable assumptions when the prompt is ambiguous and tell us what you assumed. That's part of the exercise.
If anything is broken, unclear, or blocking you from starting, let your Raft point of contact know and we'll fix it.
This repository is released under the MIT License.