Koah LabsKoah Labs

Applied Scientist

Added 6 hours ago

Who We Are

Koah Labs is building the ad network to power the next generation of AI-native products. Our mission is to help publishers monetize and help advertisers reach the right audience — without compromising speed, UX, or privacy.

We’re a small, tight-knit team in San Francisco with backgrounds at X, Apple, Meta, and early-stage startups. We’ve raised from top investors and are growing fast with real traction on both the publisher and advertiser sides.

Working at Koah means joining at the ground floor: you’ll ship code that shapes the company and the ecosystem we’re building. We move quickly, operate with high trust, and care deeply about craft.

Our Stack

  • Infra: Terraform, AWS, LGTM (Loki, Grafana, Tempo, Mimir), Tailscale, Cloudflare

  • Data: PostgreSQL, ClickHouse, Redis, Kafka, Python

  • Core Application: Ruby on Rails, React, TypeScript

  • SDKs: Flutter, React Native, Android, iOS

Example Projects

  • Design efficient algorithms for real-time bidding systems, building upon the current pricing literature

  • Create and productionize regression models to predict end conversions based on demographic, audience, and semantic data

  • Apply privacy-preserving clustering methods to categorize conversational data to improve advertiser outreach

  • Analyze and pore over data to find alpha that can improve the core ad matching system balancing publisher and advertiser outcomes

You might be a fit if

  • You have an advanced degree in Physics, Computer Science, Mathematics, Statistics, Engineering, or a related field

  • You enjoy identifying and owning challenging problems, forming testable hypotheses, and conducting impactful research to drive significant business impact

  • You have a relentless focus on continuous learning and making an impact with an ability to question the status quo

  • You have strong mathematical and statistical modeling skills

  • You enjoy communicating conclusions to both technical and non-technical audiences alike