Monitor ML

Real-time production monitoring of ML models, made simple.

3 followers

Real-time production monitoring of ML models, made simple.

3 followers

Monitor ML tracks the performance of models throughout their lifecycle and connects them to business metrics. We support model tracking, metric logging/analysis/alerting and production event logging. You choose the framework, we monitor the model.

Launch tags:Productivity•Artificial Intelligence•Tech

Launch Team

ClarifyThe autonomous CRM that helps you sell more

Promoted

Fastgen

Hey Aparna. That looks really interesting. How does Monitor ML compare to the end-to-end machine learning platforms that are currently in the market?

Report

7yr ago

Monitor ML

Maker

@mikemahlkow Thanks Mike! Monitor ML is focused on the monitoring vertical. We're platform/framework agnostic, so it doesn't matter to us how you train or deploy your models - we just want to help monitor it's performance so that you can focus on building the models. With an end-to-end platform, you're tied to their restrictions around what kind of models can be trained, deployment constraints, etc.

Report

7yr ago

This is neat. How can someone use your platform to know how their models are performing in production?

Report

7yr ago

Monitor ML

Maker

@jake_pimental1 Thanks Jake. They can use our events APIs to log model predictions. We enable human in the loop collection of ground truth so they can then define custom metrics based on what actually happened in production.

Report

7yr ago

Monitor ML

Maker

Hey everyone 👋! I'm the co-founder of Monitor ML along with my brother @eswar_dhinakaran! Our goal with Monitor ML is to bring more transparency and visibility to machine learning. ✨ From my background as a machine learning engineer, I saw models being deployed in applications with broad societal impact. However, there were very few tools to help researchers and businesses monitor models to identify how they behave over time in the real world. Monitor ML is a framework-agnostic platform to track model performance in real-time and connect models to business metrics. You can think of us as a Mixpanel/Amplitude for machine learning models. Right now we support model tracking, training metrics logging/analysis/alerting and production event logging. We support teams and individuals and work with Google Oauth (for team management) and Slack (for alerts). If you have any use cases in your company or as a developer, please email me at aparna@monitorml.com and I'm happy to chat! I'm also here to answer any questions you have :)

Report

7yr ago

Can your platform give me insights on how to improve my model?

Report

7yr ago

Monitor ML

Maker

@sumosaha Currently we have manual thresholds that set off alerts. We're working towards anomaly detection with our events platform. These insights can help drive where the model is performing abnormally. We're also integrated with Tensorflow so we're working towards scraping information from the tensorflow board while the model is training so model builders get realtime insights to when their model is moving towards/diverging from expected performance.

Report

7yr ago

Fireflies.ai 👔

This is dope. How would I upload a model's metrics quickly?

Report

7yr ago

Monitor ML

Maker

@sudotong Thanks for asking! We have a simple rest endpoint as well as our own session hook for Tensorflow sessions so you can automatically add metrics in each step of the training/evaluation phases.

Report

7yr ago