Back to core Data Science at Jellysmack after an entrepreneurship journey
I have always been curious. Not in the negative sense of the word; I mean interested in discovering and understanding how things work. This led me to study Computer Science and then do a Ph.D. in Computer Vision.
The fascinating world of academic research
My experience of lab research has been profoundly inspiring and extremely mind-stimulating. As a Ph.D. student in 2005 and then a postdoctoral researcher in 2008, I have worked in heterogeneous teams of highly skilled and motivated people on super innovative Artificial Intelligence projects in Computer Vision for mobile robotics.
Back then, our overall goal was to enable mobile robots with environmental sensing and understanding capabilities, so they could move around freely without colliding with obstacles while trying to achieve a specific task.
We were attempting to do so in real time with limited computing power, that of a laptop with a single CPU and no GPU computing, which was a challenge given the amount of data to process — to give you an idea of where we were on a technological timeline, the first iPhone was released during the second year of my PhD.
In particular, I was working on place recognition using a single camera, so a robot could realize when it was back to a spot it had previously visited. In Computer Vision, that turns out to be an image retrieval task, where the goal is to find the potential closest match to a given image from a database.
Just for the fun of it, here are a few details about how that works: the live image from a monocular camera is first encoded using the “bag of visual words” scheme. Potential locations with similar appearances are then quickly retrieved with an inverted index storing statistics about the visual words they contain. This feeds a voting system integrated into a discrete Bayesian filter that maintains a probability distribution over past locations.
In other words, each time a new image comes in, the system computes the probability of that image coming from a previously visited location. A match is declared when this probability is high for a given location.
After 6 years of research, I wanted to find out what it was like to work for a company rather than an academic lab, so I joined a company.
Back to Corsica
After the end of my postdoc in 2011, a friend from my hometown on the island of Corsica offered me to take part in an entrepreneurial project dedicated to developing modern “out of home” (OOH) advertising solutions using LED screens and interactive kiosks.
Corsica isn’t the worst location on the planet, so this would satisfy my curiosity about the workplace and my desire to return home.
I was particularly interested in that project because of its entrepreneurial side: the company was at an early stage of its life, and my role was to be a business developer.
That was the beginning of a journey of almost 10 years of different business development projects on the island, which all shared the same perspective: shine outside of Corsica and demonstrate that somehow everything is possible, even from a remote island.
Among other things during this period I have worked as export sales manager for a distillery, IT infrastructure supervisor for a drinks production and distribution company, and general manager for an e-commerce company.
I took care of tainting each of these experiences with some business intelligence and data viz, to keep programming, but more importantly to somehow keep a link with Artificial Intelligence, or more precisely, Data Science.
But over the years, I realized that I was missing what I would call “the AI research atmosphere”: addressing complicated problems for which a straightforward solution does not exist and which require head scratching + computing power to elegantly solve them.
Hurray to remote work!
Positions that would satisfy my head-scratching craving in companies close to my house did not exist until recently.
With the rise of remote work, data companies like Jellysmack have opened exciting opportunities to join teams of highly skilled data scientists and collaborate on innovative projects inside the company’s framework from home. That was the perfect combination of doing research + working in a company, + living in Corsica!
And guess what: they were looking for candidates with a Computer Vision background. I therefore applied for a job at Jellysmack and took part in several interviews including a technical one where I was asked to suggest a solution to a problem of… Image retrieval!
Believe it or not: the topic I was asked to talk about during this interview was a video subsequence matching problem that shared very strong similarities with my previous research, so I could suggest the use of bags of visual words etc…
Fortunately, the interviews went well, and I began working at Jellysmack in September of last year. I am a member of the Deep Content Understanding squad, whose purpose is to provide smart tools for video editors, and I am in charge of organizing a project called JFrame, which is about reframing a video to share it on social media using different aspect ratios.
I believe that my research experience has provided me with the AI and Computer Vision foundation required for this position, while my entrepreneurial experience has provided me with the project management skills required. It appears to me that the combination of these two is a solid match for bringing research initiatives to the reality of a marketable product.
As a result, I am pretty pleased with how things are going. Every experience is worthwhile, even if it appears to be far removed from obtained degrees or previous experience. It will eventually contribute to acquiring the expected skills for a future job — I believe what happened to me is an excellent example of this.
But another important reason that contributes to my great experience at Jellysmack is the company’s culture and its ability to unlock the potential of heterogeneous profiles, and the great news is that the company is seeking new talents: https://jobs.jellysmack.com/
