Working in Omdena vs. Working at a Tech Startup
The notion that Omdena, an online collaborative platform, is only suited for certain groups of enthusiasts is untrue, as it offers a place for nearly everyone to contribute, with the only prerequisite being passionate for the field.
In this article, we’re delighted to share some first-hand perspectives from startup professionals who’ve participated in past Omdena challenges. Erick Galinkin, Sai Praveen and Daniel Ma offer their takes in Q&A format on what’s it’s like to collaborate with others in challenges coming from a startup background including drawing similarities between the two working environments.
Enjoy! And If you’re interested in applying to the latest challenges, check out here.
Note: Views expressed by writers are purely their own and do not reflect their employers’ views.
Where do you work and what do you do?
Daniel: I work at a startup software company based in Toronto. We’re a consultancy for enterprises all throughout North America and parts of Europe. I was hired as the company’s first Data Scientist, and I specialize in the field of personalization, mainly developing recommender systems.
Erick: I’m from New York but I live out in the bay area, working at a cloud security startup. Formally, I’m a security research scientist, working on offensive security and infrastructure as a service, but my day-to-day involves tons of data and machine learning on malware and network traffic data.
Sai: I’m from Bangalore and I work for a startup company as a Data Scientist. We work with many clients to implement ML models in existing applications. My day-to-day involves data crunching, visualization, and deployment of models. I specialize in NLP.
What made you decide to contribute to Omdena?
“I always wish to do good deeds for society. I may not be in a position to contribute in terms of financial help but I heard about Omdena and their work from one of my friends.”
-Sai
Daniel: Having been more-or-less self-taught in Data Science, I was always appreciative of the community support that was offered through various mediums and from enthusiasts of all levels.
From Reddit forums to Kaggle competitions, it seemed that there was an unspoken duty for the community to help one another. I still recall the early days of getting my feet wet and posting extremely basic and borderline embarrassing questions on these channels to be delighted with thorough, encouraging responses. At the time, it’d really spurred me to overcome challenges and continue to delve into the field. For every piece of help and guidance I received, I felt the need to bank it in hopes of one day paying it forward to the community.
And that’s how I bumped into Omdena, as its mission in fostering a collaborative community strongly resonates with me.
Erick: I’ve always wanted to make a larger impact, but it can be tough for one person to really affect big changes. When I stumbled across Omdena on Reddit, I was immediately excited. The UN World Food Program is one organization whose mission I’ve always admired, and I applied immediately.
Sai: I always wish to do good deeds for society. I may not be in a position to contribute in terms of financial help but I heard about Omdena and their work from one of my friends. Omdena gave me this great opportunity to help many people and offered a place where I can contribute my skills to solve many real-world problems with the help of AI.
I applied for the UN World Food Program and learned a lot in terms of acquiring new knowledge and learning about the unique collaborative approach.
Joining Omdena is one of the best decisions I have taken in my life.
How did you get started with Omdena? Describe your role in the project?
“I was a Lead Machine Learning Engineer and worked to make our Mask R-CNN and U-Net models successful.”
-Erick
Daniel: My first project was helping Safecity, a non-profit, online platform to spread awareness of sexual violence and harassment. My role was to help develop proof-of-concept features that leverage machine learning techniques so that the organization could build upon and later implement on its existing platform. You can learn more about the work that was done here and here.
Erick: After a brief interview, I was thrilled to start working with the rest of our team on the UN WFP project. I was a Lead Machine Learning Engineer and worked to make our Mask R-CNN and U-Net models successful. We had a lot of trouble with data collection since satellite imagery can be tough to collect, especially for vulnerable parts of the world. At the end of the project, I felt that I’d offered some good mentorship, helped the project with the computer vision aspects of the problem, and although we’d not achieved as much as we’d hoped — we certainly made a dent in the problem.
Sai: In the past I was a part of the WFP, Voice4Impact projects and now I am working on Renewable Nigeria as an ML engineer. My role is not exactly defined for a particular task or role. I can join any task and make contributions, whether it’s cleaning data, making visualizations or developing models.
The WFP Project kick-started my journey at Omdena. What we needed to do was classify different crop yields seasonally throughout the year for Nepal. I’m glad I met many experts in the Image Processing field. After receiving some guidance from members, I learned how to scrape for data — unfortunately, most of the images are not helpful for the model so we had to do a lot of preprocessing and labeling of the data. The main difficult task in any real-world problem is the data, which is unstructured and there is no benchmark or standards to follow.
All the work we do will be a guide for others to research in that particular area. In the end, I felt thrilled with the results we got, even though they’re not as good as we hoped but nonetheless a great start. I’m continuing to work with Omdena, and have been doing so for the past 6 months.
What did you bring from your existing experience that helped the project and the team?
“…suggestions to keep meetings laser-focused on the agenda, limit digressions and ensure participants are equally engaged so that the group doesn’t acquiesce to ideas that come from just a few members.”
-Daniel
Daniel: Aside from going through the motions of developing data science projects (data visualization, wrangling, model development, etc), I felt I was able to exercise some of the best practices in agile project management from my day-to-day. This included suggestions to keep meetings laser-focused on the agenda, limit digressions and ensure participants are equally engaged so that the group doesn’t acquiesce to ideas that come from just a few members.
Erick: I think that the most important thing from my experience that helped was not my knowledge of computer vision or anything technical, but was my drive to start working. That is, even if the data isn’t where we want it to be, let’s start building out an architecture. Don’t worry too much just yet about tuning, let’s just get to work.
Sai: I felt like a fresher who just joined a world of big problems. In my day-to-day work, everything is sorted, clear and models will make the best inferences. At Omdena, there are many challenges hidden in one project, each step requires so many techniques and skills, along with communication between all the collaborators. My past experience helped in working with people, completing things as per schedule and keeping track of the work from all other participants. I would say I learned more from the projects than what I offered from my existing experience.
What are some similarities between volunteering at Omdena and working at your day job?
“Opinions! Whenever you have a collaborative effort, or lots of people in the room, people have lots of opinions.”
-Erick
Erick: Opinions! Whenever you have a collaborative effort or lots of people in the room, people have lots of opinions. To avoid the pitfalls of design by committee, it was important to have an “owner” for each piece of the puzzle, otherwise, we just spin our wheels discussing the pros and cons of everything, never building anything at all.
Daniel: To piggyback off Erick’s point about everyone having a perspective (which is really a good problem to have), another natural outcome from working in such an environment was that everyone’s opinions are valued. There’s not a particular reporting person the team needs to seek approval from in order to progress with the project. This is similar with startups where the organizational structure is very flat and it’s not uncommon to have engineers report to the C-suite.
When the problem is complex and many people are involved, yes, ideas can circulate and diverge endlessly. And that’s where the voluntary responsibility of the task “owner” comes in — not to pick and choose which idea should emerge, but rather to remind and set the focus of the group to deliver within the given timelines. I think most startups acknowledge that a little bit of messiness is necessary for cultivating a stimulating work environment with just the right amount of tension.
What did you take away from that could help you on the job?
“…without good, clean, high-quality data, it’s incredibly challenging to build an ML algorithm that performs well.”
-Erick
Daniel: Working with many bright individuals in the industry enhanced my ability to communicate technically; being the lone subject matter expert in my role, I’m often communicating with others at a much higher level than what’s actually going on with the work itself. Rarely, if ever do I find myself speaking with the project manager on the intricacies of precision vs. recall or the curse of dimensionality.
Erick: Some members of the UN WFP project were really clever in their data collection methods, and the things I gleaned from their work have been immensely valuable in my own work, allowing me to greatly improve my ability to quickly collect data from open sources that would have otherwise been difficult to collect manually.
It’s worth noting that this skill cannot be underestimated — without good, clean, high-quality data, it’s incredibly challenging to build an ML algorithm that performs well.
What would you say to professionals in the Data Science and AI industry that come across Omdena?
“Aside from garnering personal growth, mentorship and opportunities for application, you’ll also be staying abreast of the world’s social and environmental challenges and contributing to AI’s overall image.”
-Daniel
Erick: Pitch in! Omdena doesn’t ask much of us as individuals, and the good that can come from not only the project itself, but from the mentorship and cultivation of less experienced individuals into professionals helps the maturity, robustness, and understanding of the entire industry. The friendships alone are worth the time and effort put in.
Daniel: +1 to Erick. If your journey to Data Science and AI was anything similar to ours, you’ll understand the good that could come from a large community of active learners and enthusiasts. Aside from garnering personal growth, mentorship and opportunities for application, you’ll also be staying abreast of the world’s social and environmental challenges and contributing to AI’s overall image.
