Ruby on Rails consulting

Evil Martians have been building Rails applications since 2006, we maintain 100+ open source projects with 25 billion+ downloads, and we wrote the acclaimed book on Layered Ruby on Rails architecture. We’re the team that helps you build and grow your Ruby and Rails products, while improving team’s developer productivity. Based in New York and San Francisco, we work with startups globally.

Your Rails app needs to ship faster

You just raised a round and need to ship before the runway gets short. Or your team is small and hiring senior Rails engineers takes months. We embed directly with your team—your repo, your Slack, your standups. Not consultants who hand off a deck, but engineers who ship production code from day one. We build MVPs, APIs, billing systems, admin dashboards. We build on Rails 8 and help teams upgrade. We teach your agents to generate maintainable code. We come when needed and disconnect when done, helping you hire internal team at your own pace.

We helped StackBlitz build bolt.new’s Rails backend from early architecture through $40M+ ARR. We’ve been their engineering partner since 2021. For Lago—the YC-backed open source billing platform with $22M in funding—we are shipping product features while maintaining their high quality bar. For BetterUp—the AI coaching platform with $570M+ in funding—we built the React frontend and scalable AWS backend. For Fountain, a YC-backed startup, we shipped features and scaled infrastructure. Many of our clients are backed by Y Combinator.

Your Rails app needs to scale

The app that got you to product-market fit isn’t the app that handles 10x the traffic. Slow queries, background jobs backing up, test suites that take 30 minutes. We’ve fixed all of these.

For Whop—$150M+ in yearly payments on a Rails monolith—we cut CI time in half and prepared the zero-downtime migration of 750GB from PostgreSQL to PlanetScale. For ClickFunnels, we eliminated flaky tests and improved CI reliability from ~80% to near-100% across 9K+ unit tests. For PowerHRG, we optimized their CI pipeline. For Tines, a security automation unicorn with $90M+ raised, we delivered 100x performance improvements in target areas. For Doximity—used by 80% of US doctors—we integrated Go-powered real-time features into their Rails app.

We built the tools Rails teams reach for when things get slow. TestProf diagnoses and fixes slow test suites. Thousands of teams use it daily. AnyCable replaces Action Cable with a Go-based WebSocket server that uses 3x less memory. It powers real-time at ClickFunnels, CoinGecko, and Doximity.

Scaling Rails doesn’t mean rewriting it—it means upgrading the cookware while keeping the recipes. You keep business logic in Ruby and delegate computationally intensive work to specialized tools in Go, C, and Rust: imgproxy for image processing, AnyCable for WebSockets, serverless functions for file processing. For Playbook—a cloud platform for creatives with 700K+ users—we built serverless file processing pipelines on Google Cloud alongside their Rails app.

Your Rails architecture is holding you back

The monolith that worked at 3 engineers doesn’t work at 30. Domain boundaries are blurred, callbacks are everywhere, and new features keep breaking old ones.

Our Head of Backend, Vladimir Dementyev, wrote the highly-rated Layered Design for Ruby on Rails Applications—now in its second edition. The patterns in that book—faceless service objects replaced by form objects, workflows and operations, authorization extracted to Action Policy, notifications unified through Active Delivery, configuration made type-safe with Anyway Config—are the patterns we apply on every engagement. For the frontend layer, we use Inertia Rails with inertia_rails-contrib (built by us)—React or Vue frontends powered by Rails, no separate API. We’ve refactored codebases at Whop, Tines, and ClickFunnels without risky rewrites.

For Monograph—$30M in funding from Index Ventures and Tiger—we migrated the backend to GraphQL and moved the frontend from the monolith to Next.js.

Your Rails app needs real-time features

Chats, notifications, typing indicators, presence, cursors, collaboration, live dashboards—most Rails apps need real-time eventually. Action Cable and Solid Cable are not delivering that instant and reliable real-time UX.

We built AnyCable to solve exactly this. It keeps the familiar Action Cable API while handling connections in Go—3x less memory, delivery guarantees, resumable sessions. For Hotwire-only apps, setup is simple: swap the adapter. For ClickFunnels, we run AnyCable at scale. For Healthie, we migrated their real-time features to AnyCable. For Doximity, we built Go-powered real-time features alongside their Rails app.

You’re adding autonomous AI workflows to your Rails product

Rails is a great backend for AI products—bolt.new proved it at $40M+ ARR. We build autonomous agent pipelines in Ruby that triage issues, plan fixes, write code, self-review, open PRs, and loop on CI failures and review feedback until done. No human in the loop. Our internal agent systems run on Ruby, Roda, Sequel, and PostgreSQL with RubyLLM, checkpoint-recovery, per-task budget tracking, and CI-driven refinement.

We integrate LLM APIs, build agent infrastructure, and add AI-powered features to existing Rails applications. For real-time AI responses, we built AnyCable’s LLM streaming support—streaming LLM output over WebSockets to your frontend. We build AI Harnesses—open source skill sets that encode architectural patterns into AI coding tools like Claude Code and Codex, so agentic coding produces better Rails code. If you’re building autonomous AI workflows on Ruby, this is our territory.

Your Ruby app needs concurrency done right

Threads, Fibers, Ractors, connection pools—Ruby’s concurrency model has matured, but getting it right is still hard. Race conditions, non-atomic database transactions, and thread-safety bugs are invisible until they hit production under load.

We maintain rubocop-thread_safety (21M+ downloads) to catch concurrency bugs statically, and isolator to detect non-atomic interactions within database transactions at runtime. We made the NATS Ruby SDK 3x faster. NATS is a high-performance messaging system where concurrency correctness is everything. Our team speaks on this topic at conferences around the world: testing multithreaded code (Julia Egorova, EuRuKo 2025), threads and execution context in Ruby (Vladimir Dementyev, RubyKaigi and RubyConf AU).

Why Evil Martians for Ruby on Rails

We don’t just use Rails—we build the tools the ecosystem runs on. 25 billion+ combined downloads across 100+ open source projects: TestProf, AnyCable, Action Policy, Lefthook (1.2M weekly npm downloads), Yabeda, Logidze, isolator, anyway_config, Active Delivery, and dozens more.

We keynote at all major Ruby and Rails conferences globally. We organize the SF Ruby Conference—450+ attendees, 40+ speakers, demos from companies like bolt.new, Chime, and Gusto—and the SF Bay Area Ruby Meetup. We write books on writing maintainable sofwtare and receive awards for our open source projects. Eight of our clients grew from seed to unicorn.

Here’s who trusts us with their Rails code: bolt.new ($40M+ ARR), Whop ($150M+ yearly payments), Tines (unicorn, $90M+ raised), Doximity (80% of US doctors), BetterUp ($570M+ in funding), ClickFunnels, Lago ($22M+ raised, open source billing), Playbook ($22M+ raised, 700K+ users), Monograph ($30M, Index Ventures), Rootly (YC-backed), Fountain (YC-backed), Keygen, Synonym (a16z-backed), Chowbus, Healthie, PowerHRG, and NATS.

When to engage

There are several points where working with us makes the most sense. Right after a funding round, when there’s pressure to ship and hiring takes months. At the scaling stage, when your Rails app needs specific expertise—performance optimization, real-time infrastructure, database scaling—that doesn’t justify a full-time hire. Pre-PMF, when you need to validate fast: our design sprints have launched products in as little as two weeks.

We charge $8,000/week per engineer. We run 2–3 week sprints with clear deliverables. 65% of our clients raise their next round within 1–3 years.