Building machine learning models is hard, deploying them in production environments is even harder. If a model does not work in production, then it’s worth nothing. We all know of the famous Netflix case-study, the winning model was so complex that it was never deployed in production.
“Productionizing” machine learning requires nuanced understanding of data science, software engineering and DevOps. We too have data scientists on our team, we know that this is not an easy undertaking. Of course, we can hire but this is easier said than done.
In order to help data scientists productionize their work faster, we’re building a data science platform, rorodata. The aim is to abstract and automate all the non-data science workflows and provide a simple API to data scientists. Think of us as heroku for data scientists, on-premise and on-cloud.
Firefly, is building block in our model deployment service. This is under active development. Here we’re presenting the slides from our founder, Anand Chitipothu’s lightening talk on Firefly at Europython 2017.
Try it out and we’d love to hear your views and feedback. You can more via any of the following means:
- Visit us at http://rorodata.com
- Checkout our repo https://github.com/rorodata
- Join the discussion on slack http://slack.rorocloud.io/
Yes, we’re hiring for both full-time positions and long-term internships.
If you liked this piece, click the ❤ below so other people on Medium will see this here. Also, if you have any question or suggestion, please leave us a comment
