🎯
Declarative DSL
Compose conversation blueprints like React components, turning dataset design into a predictable, readable workflow.
Build synthetic datasets
like you build UIs
✨Declarative building blocks🧠AI-powered variations⚙️Typesafe pipelines
Like React, but for datasets — declarative and typesafe DSL for composing conversations and generating thousands of AI-powered training examples.
Compose reusable conversations with a declarative DSL and generate production-ready datasets in minutes.
Production feature set
Everything you need to ship datasets continuously
Torque borrows from the best product-engineering playbooks: versioned blueprints, strict typing, reproducible runs, and streaming observability. No YAML, no bespoke tools — just code.
🔒
Fully Typesafe
Harness Zod-powered typing with end-to-end inference so every AI training workflow stays safe, reliable, and maintainable.
🔌
Provider Agnostic
Generate with OpenAI, Anthropic, DeepSeek, vLLM, LLaMA.cpp, or any model supported by your composable data pipeline.
🤖
AI-Powered Content
Blend handcrafted prompts with AI dataset generation to produce realistic multi-message conversations at scale.
💰
Cache Optimized
Reuse context across runs to minimize token spend while keeping dataset scaling fast, efficient, and reproducible.
⚡
Concurrent Execution
Stream progress in a beautiful CLI with concurrent workers, deterministic seeds, and instant feedback loops.
The challenge
Building synthetic datasets for LLMs is tedious
Every team building AI products hits the same roadblocks when trying to create high-quality training data at scale.
📊
Sometimes you don't have enough real data
⏱️
Maintaining quality and consistency across thousands of examples is extremely time consuming
🔧
Tool calling patterns require intricate message sequences and are error‑prone
🔀
Generating different conversation flows means rewriting everything or creating various hard to maintain scripts
🎲
Designing generators that are random yet reproducible is surprisingly complex
🧩
Getting AI to understand complex composition scenarios (nested variations, conditional flows) takes significant prompt engineering time
The gaps in today's tooling
Dataset creation is still stuck in 2012-era workflows
Torque is built for product teams that expect the same polish from their AI training data as they do from their production apps. First, let's take a quick look at the pain that keeps teams from shipping.
How Torque helps
From idea to production-ready datasets in three moves
Torque replaces sprawling prompts and brittle scripts with a composable TypeScript toolkit. Every phase is automated, typed, and observable.
“Torque is the missing layer between your product engineers and the models they rely on.”
01Stage 1
Compose declarative blueprints
Describe each conversation as a reusable component. Torque keeps prompts, metadata, and policy checks together in strongly typed modules.
Ship new flows faster with versionable building blocks, not ad-hoc scripts.
02Stage 2
Scale with concurrency & caching
Spin up tens of workers, hydrate examples with AI, and cache deterministic seeds so teams get repeatable datasets without wasting tokens.
Every run logs audit trails, token spend, and policy results — plug it straight into CI.
03Stage 3
Evaluate continuously
Replay production traffic through model providers, enforce quality gates, and ship regressions straight into Slack or GitHub checks.
Close the loop with built-in evaluations and targeted retries when policies fail.
Start Building Today
Join developers building the future of LLM training data