Speak Tech Stack

AI conversational language tutor for mobile learners

Software Development San Francisco, CA 51–200 employees Privately Held

Speak builds a conversational AI language-learning app with mobile-first design (React, Swift, Kotlin) and LLM-powered tutoring (PyTorch, OpenAI). The tech stack reveals a company handling real-time user interaction (WebSockets, Redis) and training-data pipelines (Apache Airflow, BigQuery, Snowflake, dbt) — mirroring the dual challenge of scaling inference quality while managing the data infrastructure to improve model performance. Hiring velocity is decelerating while the backlog spans feature scaling, infrastructure reliability, and LLM output quality, indicating a post-launch product at growth stage.

Tech Stack 41 technologies

Core StackReact Python PyTorch Node.js TypeScript RAG Kotlin Jenkins OpenAI Kubernetes Terraform PostgreSQL Redis Apache Airflow dbt BigQuery Snowflake Redshift AWS Swift SwiftUI Android WebSockets RxJava Bitrise GCP Instagram SQL StoreKit RevenueCat+10 more

AdoptingOpenAI

What Speak Is Building

◆Challenges

Hard to access at scale
Building and maintaining data infrastructure
Improving funnel conversion
Scaling infrastructure for 10x users
Ensuring platform reliability
Improving llm performance and output quality
Increasing pipeline quality
Korean efl learner challenges
Scaling it operations
Scaling one-on-one tutoring

▲Active Projects

Launch new markets and languages
Customer success program from the ground up
Creative roadmap development
App-based video lessons
It security program
Programmatic conversational language learning product
Ml experimentation framework
Data infrastructure for training datasets
Llm-powered language learning product development
Scaling product features to many users

Hiring Activity

Decelerating60 roles · 10 in 30d

Department

Marketing

Engineering

Sales

Product

Ops

Support

Data

Content

Seniority

Mid

Senior

Manager

Lead

Junior

Director

Staff

Notable leadership hires: Growth Head, Sales Development Lead, Head of Growth

Company intelligence

Find more companies like Speak by tech stack, pain points and active projects

Get started free

About Speak

Speak is a mobile language-learning platform powered by conversational AI, launched in 2019 and operating across 15+ languages. The product targets two billion people globally attempting language acquisition, positioning one-on-one tutoring at scale as the core value proposition. The company operates a distributed team across San Francisco, Seoul, Tokyo, Taipei, and Ljubljana, with a 51–200 person headcount split across marketing, engineering, sales, product, and data functions. Current priorities center on geographic expansion (new markets and languages), customer success infrastructure, app-based video lessons, and scaling the platform to handle 10x user growth while maintaining LLM output quality.

HeadquartersSan Francisco, CA

Company Size51–200 employees

Hiring MarketsUnited States, South Korea, Japan, Taiwan

Frequently Asked Questions

What tech stack does Speak use for AI language learning?

Speak runs React and Node.js for web, Swift/SwiftUI and Kotlin/RxJava for mobile, PyTorch for model inference, and OpenAI for LLM capabilities. Backend uses PostgreSQL, Redis, Kubernetes, and GCP/AWS. Data pipelines rely on Apache Airflow, dbt, BigQuery, and Snowflake.

Where is Speak headquartered?

Speak is headquartered in San Francisco, CA, with distributed offices in Seoul, Tokyo, Taipei, and Ljubljana. The company employs 51–200 people across these locations.