Last week we hosted our sixth “Data Science, Machine Learning and Kubeflow” Meetup. Special thanks to our awesome speakers Kausthub Krishnamurthy and Mamta Shukla. In this blog post we’ll recap some highlights from the Meetup and preview what’s next. Ok, let’s dig in.
Join a Meetup near you
First, if you missed last week’s Meetup? No need to suffer from FOMO. Here’s a list of the Meetups that are part of the “Data Science, Machine Learning and Kubeflow” Meetup network. Please join the one that is the most time friendly to your location.
- New York
- San Francisco
- Silicon Valley
Get involved in the Kubeflow community
- Join Kubeflow Community Slack
- Are you interested in speaking at a future Meetup?
- Is your company interested in sponsoring a Meetup?
- Would you like to be a co-organizer of a local Meetup?
If you answered yes to any of the above, Send one of the organizers/hosts a message on Meetup.com or jump onto Kubeflow Community Slack and DM @rawkintrevo
Charitable Giving on Behalf of the Meetup
With the unprecedented circumstances facing our global community, Arrikto is looking for even more ways to contribute. Normally we vote on this- but Trevor forgot to hold a vote since he’s new, but we picked UNHCR for the support of Ukrainian refugees. UNHCR, the UN Refugee Agency, is a global organization working to save lives, protect rights and build a better future for refugees, internally displaced communities and stateless people. This charity helps ensure that Ukrainians forced to flee their homes are sheltered and safe. We are pleased to be making a donation of $100 to them on behalf of the Kubeflow community. Again, thanks to all of you who attended and voted!
Talk #1: Deep Learning in Robotic Vision – A Confluence of Architectures
The confluence of Deep Learning being applied to Robotics is the nexus of rapidly evolving technologies, both for industrial automation as well as field robotics. This talk gives us an introduction into the technological opportunities, and the architectural challenges that come with operating in the crossroads of these fields as they transition from labs and universities and into scalable real-world applications.
Kausthub Krishnamurthy is a Machine Learning & Robotics Engineer, with a focus on Computer Vision and Sensing Technologies. In both industrial robotics and mobile robotic environments, he has applied deep learning solutions to create proof of concept models, and is currently looking at leveraging Kubeflow to explore scalable deployed Deep Learning solutions.
Talk #2: Porting Signal Processing Algorithms to CuPy for Precision Measurement
At CERN, for the alignment of large superconducting magnets and cryogenics, an interferometry based system is being devised to identify the position of their elements. The data acquired from photo-detection module, received after a sweep of laser source, needs to undergo post processing to obtain the final results. The system must monitor position of a large number of elements every second. Thus, GPU was employed to provide faster and precise results. This required to use signal processing algorithms like: Butterworth Filter, Hilbert Transform, Savitzky-Golay smoothing Filter in GPU. This talk will cover steps involved in adopting signal processing algorithm to GPU to achieve better performance and highlight performance metrics with respect to increase in the data size and possible optimizations of its processing.
Mamta Shukla graduated as an electronics and telecommunication engineer from India in 2018. She started her journey in tech as an Outreachy Intern and worked with Linux GPU subsystem and contributed to VKMS driver. She is an open source enthusiast and loves to contribute and promote FOSS. Later she worked as a Fellow in Beams Department at European Organization for Nuclear Research(CERN), Geneva, Switzerland and developed device drivers for in-house electronics. At CERN, she worked on Frequency Scanning Interferometry (FSI) system which will be used for alignment of Cryogenics for High-Luminosiy LHC. While working for FSI, with experience of GPU subsystem – developed and ported signal processing algorithms on GPU using CuPy.
There was also one short lightning talk at the Meetup worth checking out.
- A 10 Minute Introduction to Kubeflow: Basics, Architecture & Components – Jimmy Guerrero, VP Developer Relations (Arrikto)
Questions and Answers
Here’s a recap of some of the Q&A during the Meetup edited for brevity and readability.
An attendee asked Kausthub, “Did you work on ML models to coordinate the behavior of multiple robots?”
Yes! It was crazy interesting trying to coordinate multiple aerial drones to map out an area in a collaborative map building manner.
Though the ML wasn’t applied specifically to the coordination or map combining problem – there are other incredibly intelligent mathematical methods we can use for that (way above my capacity most of the time).
Where ML comes into something like that is in a few ways (and many more i’ve never seen):
1. identifying critical landmarks
2. identifying semantic information (e.g. one robot may be able to map out all water areas for no-landing zones so the other drones can plan more safely). They can share specialized information so that asymmetric multi-agent robot systems can work more efficiently.
3. Comms Authentication – how do I, as a robot, know the data coming in is from a friendly robot on my team, and not from some other nefarious actor feeding me false information. Identifying spoofing patterns in signals is another area where ML may apply in a big way.
4. data compression & encryption – the pointcloud collected by one robot might be super heavy to send – using an encoder model on the transmitter robot, and a match-trained decoder may help improve compression in a way that can help us optimize the comms and even protect the encryption of the data in transit.
I am new to this meetup, what if I want to contribute?
Awesome! We’d love to have you! There are multiple ways to get involved.
Getting involved with a community can be an intimidating thing to do, but we were all just getting started once, so don’t be nervous!
Great ways to get started with meetups are presenting content- in this meetup we love all things DS/ML/Statistics/AI and MLOps, so you have a wide field to choose from. Another great way to get involved is to be a local meetup co-host. As events start happening in person this can be a huge way to become a centerpiece in your local community.
If you’re interested in either of these options- please reach out to email@example.com.
Upcoming April 2022 Meetup
We are excited to announce that we have our speakers locked in for the next meetup.
April 7, 2022
- Elastic & Automated Time Series Predictive Modeling – Yuhui (Yoshi) Shi
- Introducing the Potluck ML Framework – KUNGFU.AI
If you are new to Kubeflow – install MiniKF
MIniKF is the easiest way to get started with Kubeflow on the platform of your choice (AWS or GCP.)
Here’s the links:
Get started with Kubeflow – hands-on tutorials
Installed but don’t know where to start? Get started with these hands-on, practical Kubeflow tutorials.
- Tutorial 1: An End-to-End ML Workflow: From Notebook to Kubeflow Pipelines with MiniKF & Kale
- Tutorial 2: Build An End-to-End ML Workflow: From Notebook to HP Tuning to Kubeflow Pipelines with Kale
- Tutorial 3: Build an ML pipeline with hyperparameter tuning and serve the model starting from a notebook
- Tutorial 4: Build an AutoML workflow starting from a notebook
- Tutorial 5: Distributed Training on Kubernetes with Kubeflow, Kale and PyTorch
FREE Kubeflow courses and certifications
We are excited to announce the first of several free instructor-led and on-demand Kubeflow courses! The “Introduction to Kubeflow” series of courses will start with the fundamentals, then go on to deeper dives of various Kubeflow components. Each course will be delivered over Zoom with the opportunity to earn a certificate upon successful completion of an exam. Visit us to learn more.
We hope to see you at a future Meetup!