Case Study: Unbabel achieves over 50% cost savings and halves latency with Amazon Web Services

Unbabel - Customer Case Study

Unbabel is an AI-plus-human translation platform that enables fast, low‑cost localization for businesses and manages a community of more than 40,000 translators. Early in its growth the company faced frequent downtime, poor scaling, and rising infrastructure costs (previous bills exceeded $12,000/month) because its initial cloud provider didn’t offer the control or services needed for its NLP and machine‑learning workloads.

Unbabel migrated its core stack to AWS (using EC2 with Auto Scaling and ELB, Amazon VPC, RDS for PostgreSQL, S3, CloudWatch, IAM and Amazon Elasticsearch Service) and optimized worker processes. The move cut infrastructure bills by more than 50%, halved average web response times (250 → 125 ms), eliminated deployment downtime, reduced ops effort to about 0.25 FTE, and freed resources to launch new services and extend AWS use further.

Unbabel

João Graça

Chief Technology Officer