Suggestions
Eero Laaksonen
CEO at Valohai – The last MLOps platform you will try.
Eero Laaksonen is the CEO and founder of Valohai, a company that provides a scalable MLOps platform.13 He describes himself as a "serial entrepreneur hippie" with a keen interest in making the world a better place and believes people should work less and enjoy life more.14
Eero founded Valohai in October 2016, focusing on automating and streamlining machine learning workflows for enterprises.12 The company aims to cut engineering workload related to large-scale machine learning by automating, versioning, and tracking ML pipelines from feature extraction to model deployment.1
Prior to Valohai, Eero held positions such as:
- Founder of Turku.AI Meetup (October 2016 - Present)
- Member of the Board at Verkkoteollisuus ry (eCommerce Finland) (January 2016 - Present)
- Solutions Architect at Ixonos (February 2016 - August 2016)
- Senior Advisor and Senior Partner at Anders Innovations (January 2008 - February 2016)1
Eero is based in San Francisco, California, and can be contacted at [email protected] or +358 40 545 9991.13
Highlights
Hey, I’ve got an idea 💡
Let’s stop using Slack and build our own communications platform.
Starting today, you allocate all engineering resources to your DIY platform. In the meantime, other functions can’t get their work done because the platform isn’t ready.
This sounds ridiculous, doesn’t it?
But this is exactly what we’ve seen over a decade in the ML space: Many teams build their own ML platforms from scratch.
The problem is that these DIY platforms don’t generate value or ROI by mere existence. The value comes from having ML models in production.
(Bonus points for doing so reliably and in less time.)
But it only gets worse: You’ll need to build and maintain every new feature, which means hiring and retaining the right engineering talent.
And when you start to scale, your DIY platform needs to ensure fast onboarding times for new hires, such as subject matter experts with limited engineering skills.
I can go on, but I hope you get the point.
So tell me this:
Why do we use Slack or Teams for internal communications, HubSpot or Salesforce for most of the sales and marketing work, but we build our own ML platforms? 🧐
The Data Hairball 😱
It's a topic that comes up again and again in calls we have with folks working on deep learning, especially computer vision.
This week, Juha did a fun presentation on how Valohai's new dataset feature helps solve versioning and managing datasets. Tracking datasets in the same place where you do preprocessing is such a no-brainer.
More 👉 https://lnkd.in/dYBQnyHj
#deeplearning #computervision #mlops
