Scaling Localization with AI at Lyft

Summary

Lyft re-architected its batch localization pipeline to combine artificial intelligence with professional linguist oversight. The previous workflow relied exclusively on human translation, which resulted in multi-day turnarounds and linear cost scaling that hindered rapid market expansions. Under the new architecture, a dual-path system generates three candidate translations using a fast generation model, evaluates them against defined rubrics using a reasoning-focused model, and immediately releases approved drafts. Linguists review the drafts asynchronously in the Smartling translation management system to establish the final system of record. This pipeline reduced delivery latency from days to minutes while achieving a state where 95% of machine-generated translations require no significant changes during human review.

Context

Lyft's expansion goals, including compliance with Québec's Bill 96 and European growth requiring six new languages, outpaced its traditional localization model, which relied entirely on human translation and scaled linearly with multi-day turnaround times.

Approach / What changed

Lyft implemented a dual-path batch pipeline using a Drafter model to generate three candidates using UI context, an Evaluator reasoning model to score and select optimal outputs for early release, and asynchronous linguist review in the Smartling TMS. The system incorporates Pydantic schemas, locale-specific prompt overrides, traffic splitting, and configuration-driven model rollouts.

Takeaways

• Lyft uses a fast generative model (such as Claude Haiku or GPT mini models) for candidate drafting, combined with a reasoning-focused model to evaluate translation quality and semantic drift.
• Configuring specific prompt overrides for closely related language variants (like en-GB and en-CA) restricts transformations to orthographic and spelling updates, preventing unwanted full rewrites.
• Approximately 95% of early-released AI translations require no significant changes during subsequent linguist review, with human oversight focusing on complex idioms and legal disclaimers.

Related reading

• Amazon · May 14, 2026

  Promptimus: Improving already good LLM prompts with zero manual engineering

  Enterprise applications rely on mature large language model prompts that incorporate intricate business logic and regulatory constraints, making further optimization and cross-model migration challenging. Promptimus automates prompt refinement through a four-step iterative loop that evaluates candidate prompts against decomposed metric checkpoints, diagnoses failure bottlenecks, and generates targeted improvement strategies. The system operates either via full prompt rewrites in standard mode or through surgical, programmatic find-and-replace edits in edit mode to preserve structured taxonomies and schemas. Tested across multimodal classification and code generation tasks, the framework achieves performance gains with small development datasets of 20 to 50 samples. Amazon is making the model-agnostic optimization system available through Amazon Bedrock to support enterprise generative-AI model migration without manual engineering.

  Zhengyuan Shen, Yunfei Bai, Sullam Jeoung, Shuai Wang

• Lyft