Speak builds conversational AI tutors using OpenAI, with a tech stack anchored on PyTorch, RAG, BigQuery, and Snowflake for LLM training and data pipelines. The company is actively scaling across multiple markets and languages while managing core challenges around LLM output quality, content hygiene, and platform reliability. Hiring velocity is accelerating with engineering and marketing leading growth — a pattern consistent with simultaneous product expansion (new markets, B2B entry, app redesign) and distribution challenges in competitive language-learning segments.
Tech Stack 39 technologies
What Speak Is Building
◆Challenges
- Hard to access at scale
- Ensuring platform reliability
- Expanding reach in japan
- Improving llm performance and output quality
- Scaling one-on-one tutoring
- Improving language learning at scale
- Scaling content
- Output velocity
- Data hygiene issues
- Scaling product features to many users
▲Active Projects
- Launch new markets and languages
- App-based video lessons
- S4b b2b expansion
- Programmatic conversational language learning product
- Performance optimization for korean b2c
- Data infrastructure for training datasets
- Llm-powered language learning product development
- Scaling product features to many users
- Refining llm app building process
- Ai tutoring product redesign
Hiring Activity
45 roles · 15 in 30d
Department
Seniority
Notable leadership hires: Head of Growth, Head of Marketing
Company intelligence