November 2025
Selected for Amazon AWS Startups Programme
Doses AI has been selected for the Amazon Web Services Activate programme for AI startups. The programme provides AWS credits, technical support, and business mentorship — giving Doses AI cloud infrastructure to complement its edge-focused deployment strategy.
About AWS Activate
AWS Activate is Amazon's flagship programme for startups, providing cloud credits, technical guidance, and business support to help early-stage companies build and scale on AWS infrastructure. The programme has supported hundreds of thousands of startups globally, offering a structured path from prototype through to production-grade cloud architecture.
While Doses AI's products run entirely on-device — with no user data ever leaving the smartphone — the research and training infrastructure behind those models is cloud-native. Training a ternary language model requires distributed GPU clusters, large-scale data pipelines, and experiment tracking systems that benefit from the elasticity and tooling of modern cloud platforms. AWS provides the backend that makes the edge deployment possible.
AWS Activate provides startups with up to $100K in AWS credits, technical support from AWS solution architects, and access to the AWS Partner Network. The programme supports startups across all stages, from pre-seed companies building their first prototype to growth-stage companies scaling production workloads.
Cloud Meets Edge
For Doses AI, the AWS infrastructure addresses the training side of their development pipeline while their products remain fully on-device:
- SageMaker training jobs — Managed training infrastructure for running distributed model training experiments, with built-in experiment tracking, hyperparameter tuning, and automatic model versioning
- EC2 GPU instances — On-demand access to P4d and P5 instances for large-scale training runs, supporting the compute-intensive continual pretraining stage of the distillation pipeline
- S3 data pipelines — Scalable storage and data processing infrastructure for managing training datasets, model checkpoints, and evaluation benchmarks across the full model lineup
- Technical architecture support — Access to AWS solution architects specialising in ML infrastructure, helping optimise distributed training configurations and reduce per-experiment costs