Sai Kumar M.K.
Sr. Applied Scientist · Canva Video Studio
35 posts
15 projects
2.5 yrs @ Canva
Types
MLHPCResearchMathsReverse Eng
Badge case
005
006
007
008
Machine Learning Engineer | Research Engineer | Applied Scientist working on generative AI in video at Canva
I'm a senior applied scientist at Canva in the video space, having graduated from Monash aon my thesis on "Bias Modelling Mitigation in Diffusion Models". I was also the former president of the Monash Association of Coding (MAC), where we saw record growth to over 1000 members, more significant flagship events such as our Tech Careers Evening and MACathon, mock interviews, and more frequent events.
As for what I do? I do it all, my story started in reverse engineering, but I studied applied mathematics in university, then worked in computer vision in the photo and video space. I'm a research engineer at heart, building training infrastructure for large-scale problems that Canva has for improving their product experience for users, but I also work extensively on making models go brr - I love the idea of making services faster, as I often end up upgrading services incidentally!
If you have any questions about my experiences at Canva or want to explore my personal projects and blog pieces, feel free to reach out. You can contact me via email (linked in my resume), on LinkedIn using my full name, or find me on Discord at Theorvolt. Also, remember to check out my GitHub. Happy reading!
Over the summer of 2023, I had the unique opportunity of doing a Machine Learning Engineering internship at Canva. The project involved researching AI safety techniques for AI image generation, and unlike an ordinary project, my objective was to present my findings by the end of the internship. My project was more akin to industrial research and development (R&D), as there was no signposting for my project, only the discoveries I made. Industry research roles fall under titles like Applied Scientist, Machine Learning Engineer, or Research Engineer. Industry research deviates from research in Academia, as you'll ultimately be researching techniques and results that contribute towards a product; in the case of Canva, an online design platform, it would be relating to image generation, or more generally; to design.
Adobe is in a similar space, where they employ research engineers and applied scientists, who work on applications such as Adobe Firefly (a generative AI application relating to photo editing and generation). In general terms, an Applied Scientist works on researching, experimenting, and presenting proof of concept ideas, and if given the go-ahead, a Machine Learning Engineer will then work to productionise that idea. A Research engineer will do some amount of experimental and production work.
Openings for these roles are scarce, with very few openings for undergraduates as they look to hire Masters by Research or PhD students who have either a catalogue of published work or expertise on a topic that won't be found in any undergraduate curriculum. An honours year is a great start for these roles, as you'll be synthesising a piece on a topic you'd have researched and experimented with.
In academia, you have more control of the research topic, while the driving force of industrial R&D is that somewhere down the line, the decision to invest in research yields a profit. There is still an aspect of freedom in what you're able to research, but with the expectation that you produce novel results that can be turned into a product. Some other examples of industrial research include Google's DeepMind, which has developed some of its large-scale AI but has considerable research output, and OpenAI, whose premise involves rapid deployment of discoveries across text, image, audio, and so on.
Compared to high school, your actions have much more weight and are more noticeable. If you want to make the most of university, attend events outside of classes, and don't focus exclusively on your grades, you gain little if you spend all your time on it. Student clubs and teams are the highlight of university, you can learn much from everyone around you. I got introduced to my former club by talking to the former president about resumes, and the effect of having a more experienced community to look over your shoulders and guide you is immense, so make many friends.
When it comes to opportunities, whether they're internships, graduate roles, or research experiences, be proactive, if you're hesitant about applying somewhere, just do it, and do it early, and work alongside your peers to help each other with these opportunities. Even if you don't feel you'd get selected for an opportunity, the practice and experience gained from applying are invaluable and help for future applications.
Additionally, use university courses not to make yourself more employable, but to study courses that are intellectually stimulating or interesting for you; you have a lot of time outside of class to develop yourself as a candidate. But most importantly, don't rush university, you'll have time to learn more about what you want to do with the countless opportunities presented.
Check out my Grad/Intern Playbook for comprehensive guidance on landing your first tech role.
Have questions about my experiences at Canva, want to discuss ML research, or looking for career advice? Feel free to reach out!
Sr. Applied Scientist · Canva Video Studio
Types
Badge case
Email available in résumé
LinkedIn: Search "Sai Kumar Monash"
Discord: Theorvolt