Senren:
The Future of
ML Infrastructure
ML Infrastructure
Blog (Coming soon)
Join the waitlist
Feature store
End-to-end feature platform combining low-latency serving and time-travel capabilities - accelerate ML development with reproducible features.
Get to production 10x faster with Senren ML Platform
We build tools in Rust to provide the best possible DevEx with no compromises on capabilities or performance.
04
02
Advanced Routing
Deploy new versions with confidence. Our features include automated A/B testing, custom traffic splitting, and shadow testing for zero-risk deployments.
Built for Scale
Cloud-native infrastructure with intelligent data sharding, multi-region deployment, and comprehensive observability - built to scale with your ML operations.
05
Model Serving
Seamlessly serve models built in TensorFlow, PyTorch, or pure Python. Our intelligent infrastructure adapts to your stack.
03
Serving API
Our API-first platform transforms your models into production-ready services with minimal configuration required.
01
senren apply
Deploy to production:
senren apply
Deploy to
production:
production:
senren init <dir>
define your infrastructure in python:
pip install senren
Install python SDK:
senren apply
Deploy to production:
senren init <dir>
define your infrastructure in python:
pip install senren
Install python SDK:
How it works
manage all Kinds of Features
- Request: declare features which are a part of the RequestParams and log them for later use.
- Real-time: compute new features on-the-fly using powerful SQL-like expressions.
- Batch: efficiently process historical data
- Streaming: generate features from your event streams, ensuring that your models always have the latest signals.
Senren's Feature Store handles the full cycle of feature development:
Deploy Any Model
- Production-Ready Infrastructure: Get automatic scaling, monitoring, and high availability without DevOps overhead.
- GPU & CPU Support: Choose the right hardware for your workload, from cost-effective CPU inference to high-performance GPU acceleration.
- Custom Python Code: Run your specialized algorithms and business logic with full Python flexibility. If it runs in Python, it runs on Senren.
- Framework-based Models: Deploy models built with PyTorch, TensorFlow and Triton.
Senren's ML Platform adapts to your workflow, not the other way around. We support:
One Endpoint for everything
- End-to-End Tracking: Automatically capture and store all predictions and model behaviors with built-in tracking configurations.
- Production-Grade Monitoring: Get comprehensive monitoring out of the box for every model variant.
- Feature Service Integration: Seamlessly connect your models with feature computation for real-time inference.
- Shadow Deployments: Test new models in production with zero risk by processing real traffic without affecting user experience.
- Sophisticated Traffic Management: Easily control traffic splitting between model variants for A/B testing and gradual rollouts.
Ship faster and experiment with confidence using Senren's powerful serving abstraction. A single endpoint gives you:
Model Serving
Whether a framework or a custom python code — both could be deployed on Senren ML Platform.
Feature Store
Batch, streaming, request- and real-time features out of the box.
Managed Databases
Choose the right storage for your use-case and optimize costs
Streamline deployments with Senren
Senren Python SDK provides just the right abstractions to make the deployment of your models a breeze.
Advanced Routing capabilities allow you to run both A/B tests and Shadow tests and be confident in your new deployment.
Monitoring and Tracking come out of the box. Flexible configurations allow for custom alerting rules.
ads = ServingEndpoint(
host="ads.senren.io",
header="x-senren",
routing={
HeaderValue("control"): ModelConfig(
traffic_share=80.0,
feature_svc=ads_control_fs,
model_svc=ads_control_ms,
default=True
),
HeaderValue("challenger"): ModelConfig(
traffic_share=20.0,
feature_svc=ads_challenger_fs,
model_svc=ads_challenger_ms,
default=False,
).attach(
ModelShadowEndpoint(
traffic_share=100.0,
feature_svc=ads_challenger_v2_fs,
model_svc=ads_challenger_v2_ms,
)
)
}
)
One endpoint
Routing, fetching features with real-time calculation and applying models in one call.
We're launching soon!
Join the waitlist
© All Rights Reserved. Senren Inc.
hello@senren.io
