Superb AI is a global company with home bases in the USA and Korea. Our community team in both regions creates content and shares compelling resources about computer vision, MLOps and related data operations with a community of practitioners and industry leaders.
Here, we would like to introduce team members and show how they go about their mission of supporting the adoption of AI.
• Dasha Gurova - Community Manager based in San Diego. Dasha curates The Ground Truth, an essential newsletter for computer vision practitioners.
• James Le - Data Advocate based in San Francisco. James is the host of Datacast, a podcast that follows the career journeys of data practitioners and researchers and unpacks the lessons they learned along the way.
• Olivia Ha - Community Manager based in Korea. Olivia curates MLOps Insight, a newsletter for those interested in MLOps and computer vision.
Questions for Dasha Gurova
1 - Please introduce yourself
My name is Dasha Gurova and I joined Superb AI fairly recently in September 2021. The way I see it, a community manager’s work is to be a connector and enabler of people interested in the same topic. Therefore, my work is to represent the interests of the computer vision community at Superb AI and communicate Superb AI’s vision and solutions back to the community.
I have a few initiatives in the works, one of which is The Ground Truth - a newsletter where I curate news and useful resources for computer vision practitioners with a focus on data-centric practices. Besides that, I am contributing to the content we publish on our website. My future plans include hosting community events and working on technical computer vision tutorials.
2 - What’s your favorite part about writing your newsletters?
I believe that knowledge-sharing is the single most effective way to promote innovation. And newsletters are one of the best ways to do that. There are a few things I find beneficial in writing one:
- It’s a great incentive for me to never stop learning.
- It's a creative process - I always push myself to make each newsletter issue more interesting, storylike and fun.
- The newsletter is my contribution to the adoption of computer vision and my way of creating value for a community of practitioners. That, to me,t is meaningful.
3 - What do you think of the MLOps market?
Let me start with a quote here: “Everybody gangsta until real-world deployment in production.” (Andrej Karpathy)
To take AI projects into production, we need tools and frameworks similar to DevOps. MLOps is still a new, but very fast developing ecosystem, and I believe that it will have profound effects on the way we develop and ship AI systems. A robust MLOps stack is necessary for broader adoption of AI, as it will make it easier for smaller teams and non-tech companies to integrate AI into their products.
4 - What is your difficulty in creating a newsletter?
Artificial Intelligence is an incredibly hyped-up topic these days, and everyone is trying to generate traffic from it. As a result, there are many non-informative articles or misleading claims written just to use those keywords. Moreover, since ML in production is not mature yet, practical knowledge on MLOps is scarce and not easy to find. This makes finding useful and credible resources somewhat challenging, but important work.
5 - What does the process of writing your newsletters look like for you?
I read a lot of resources, watch webinars, listen to podcasts, and look out for the news and fresh papers. As I do that I pick and choose content relevant for computer vision practitioners and data scientists. With each resource or article that catches my eye, I ask myself: “Would I find this useful/interesting?”
Another part of it is that I think of The Ground Truth as a conversation with computer vision practitioners and learners. This makes it personal and when I write the newsletter I always try to make it more story-like and conversational.
6 - What other content do you like consuming?
I love newsletters and subscribe to many on various topics. I find them very useful because as long as I trust the author, all news and helpful resources show up in my inbox regularly.
I enjoy podcasts because usually, guests are sharing really valuable experiences, rather than theory. I am also a big fan of content repositories maintained by practitioners or communities; you can think of it as tribal knowledge obtained in action. I am currently working on a repository with all the resources I like and will share as soon as possible through my newsletter. Stay tuned.
Subscribe to The Ground Truth Newsletter
Questions for James Le
1 - Please introduce yourself
My name is James Le and I’m a Data Advocate at Superb AI. Since joining the company in March 2021, I have been responsible for driving product-led growth via content and partnerships.
From the content side, I write thought leadership and product-focused articles to educate Superb AI’s blog readers about our industry and our product offerings. From the partnerships side, I serve as the liaison between Superb AI and other startup partners in the AI Infrastructure Alliance and facilitate joint marketing and product integration initiatives. I also collaborate closely with Dasha on community initiatives in the US, such as finding meetups to go to and conferences to speak at.
Outside of my responsibilities at Superb AI, I’m heavily involved with the broad data and ML communities via my personal blog and podcast.
2 - What motivated you to start your podcast?
I started Datacast back in August 2018, when I was about to enroll into graduate school to get my Master’s degree in Computer Science. This happened during a period in which I attempted to consume as much knowledge about data science as possible. I figured that having conversations with data practitioners and learning about their professional journey would inspire me to craft my own journey.
Initially, I talked mainly with data scientists and ML researchers. As my knowledge of the field evolves, the personas of podcast guests also diversify data analysts, data engineers, academic professors, Ph.D. students, etc. These days, I mostly interview data/ML founders and investors to equip my knowledge about the data and ML infrastructure ecosystems - considering that Superb AI is building tools that belong to these spaces.
3 - How do you grow your audience?
I actually do not pay too much attention to growing my podcast audience at the beginning. I spent a lot of time on finding good guests, conducting in-depth research about them, and crafting excellent questions. I truly believe that my curiosity engine and personal touch to the interviews resonate with the audience.
My goal with the podcast is to feature raw conversations with data and ML practitioners and researchers to unpack the narrative journeys of their careers. Hopefully, these conversations can inspire early-stage folks who are navigating their own path in this exciting space. Growth happens organically as a result of the value that listeners get.
4 - Who was the most impressed guest so far?
It’s challenging to pick just one guest, so I’ll provide three episodes for you to listen to:
1. The episode with Sara Hooker if you are a scientist or researcher.
2. The episode with Prukalpa Sankar if you are a founder.
3. The episode with Luigi Patruno if you are an engineer.
5 - What goals do you have for the future?
My goals with Superb AI for 2022 are to continue creating good written content, exploring more fruitful partnerships, and speaking at more industry conferences. For the podcast, I hope to release the 100th episode by the end of the year.
6 - What are some of your favorite newsletters, books, and podcasts?
Last summer, I collaborated with a friend to build a living site for data resources in Notion. The range of resources (newsletters, communities, podcasts, and more) should provide learning mechanisms suitable for data-inquisitive minds at all stages of their journey. Check it out at this link!
Questions for Olivia Ha
1 - Please introduce yourself
My name is Olivia and I've been working at Superb AI for the past year. Hired the first community growth manager at Superb AI, I have been a bridge between the community and Superb AI. One highlight was the first MLOps community conference in April in Korea which I organized with other MLOps KR organizers. That was the launchpad for Superb AI's growing presence in the practitioners' community. In addition to my job as a community growth manager, I also contribute to creating content that reaches our community members through newsletters, webinars, blog posts, and various other campaigns. The best part about my role at Superb AI is that I have various collaboration opportunities across teams every single day and keep updated about the product, customers and the industry. I previously worked as an MVP influencer community associate at Microsoft, where I empowered and supported developer influencers in the Asia Pacific region. Outside of work, I have initiated several community events to empower Women in Tech and practitioners in the data world.
2 - What are your thoughts about MLOps?
There are many ways to define MLOps but I would like to quote Jamal Robinson (Head of AI/ML Business Development at Amazon)'s the definition. 'MLOps is best practices and policies for businesses to successfully use machine learning in an explainable, repeatable and production-ready manner'.
MLOps was brand new to me when I first joined and I still think that the best way to understand this rapidly developing field is to just get your hands dirty. Publishing MLOps Insight has helped me to get a good grasp on the landscape of MLOps. The MLOps practice is growing rapidly in both academic and practical circles, in fact, in Korea, where I'm one of the organizers, the community is almost 3k members strong.
I would argue that MLOps will be indispensable for any organization that are producing AI models and collecting real-time data. Unless companies can leverage real-time data to retrain their models and provide unique value to their customers, they will lose their competitive edge. If there's one key takeaway for me, it's that data management will be one of the most important aspects of MLOps.
3 - What does the process of writing your newsletters look like for you?
The newsletter is published on the third Monday of every month. I'm the only person responsible for curation, and because the newsletter comes out infrequently, it's a big job to select relevant and interesting content from the past month. So the first step is to start putting each issue's contents together one week before publishing. To bring fresh content that's not corny, I refer to lots of sources including different global corporations’ newsletters, engineers’ global communities, industry leaders’ tweets and blogs, etc. My favorite source is actually our team Slack channels called read-ml, read-news, where our engineers and developers share interesting news. The newsletter has only been issued six times, and we're still experimenting with our design, content, and publication timeline. Recently, for example, we transitioned to including long-form journalism content, and I have the opportunity to pick out the information that's most topical for our readers to hear about. I’ve also tried to add the feedback button which our readers could feel closer to us and share their opinion with us.
4 - What is the most challenging part of making the newsletter?
Sometimes it’s overwhelming to look through so much content, and sometimes the language in our sources can be very technical. I don’t think this will ever get easier. (haha) A big part of the editing process is how our readers perceive our newsletter. I'm so impressed by our intelligent and thoughtful readership, and it makes me want to deliver a high standard of content for them. I do hope that our readers find benefits and insight from our newsletter. I added a feedback button to our most recent issue, and we're always opening up more opportunities to communicate with our readers and tailor our content to their tastes.
5 - What value subscribers can find from MLOps Insight?
We’ve started MLOps insight because we want to help our readers find resources about MLOps from data-centric perspectives more easily. It can be difficult to find Korean voices that highlight data-centric MLOps and we would like to motivate practitioners who may be struggling within this field. We would also like to advise decision-makers who underestimate the importance of data in AI. In addition to this, we are writing this newsletter from our experiences in the field and we plan on making this a dynamic newsletter open to readers’ feedback. We’d like to seek more ways to communicate with our readers and we strive to create content that is applicable to our readers.