I interned at Artifacia for a period of 2 months from December 2015 to January 2016. The learning and experience I gained there was exponential!
I applied for an internship at Artifacia with a Machine Learning and Data Science background. I had just started to get a hang of the newer Deep Learning algorithms. Artifacia was already one of the top startups in India in the field of Visual Intelligence. I got selected for the Research Intern role after a couple of interviews, a Kaggle round, and a programming round.
Most of my work in the first month of the internship required me to work on applied research problems in their newly started R&D division named Artifacia Research. We worked on a bunch of problems like Multimedia Sentiment Analysis(Image+Text), Image Captioning, Visual Question Answering, etc. The most interesting part about all this is the fact that we were able to complete the first version of each one of these projects within a single week! I definitely have to attribute this to Navneet (CEO and Co-Founder) because working on research problems at such high throughput needs a good leader to steer the group in the right direction. Another important quality in the research team back then was that there was a great synergy between the skill sets that I had which was mostly in NLP, and the company's Co-Founder and CTO Vivek who was leading Artifacia Research with specialization in Computer Vision. With this particular combination of skills, it was easy to work on problems at the intersection of Computer Vision and NLP. We had meetings every day, sometimes multiple times in the day to constantly assess the status of the project and brainstorming ideas quickly to maintain the momentum of our project. Another interesting part is that Navneet never seemed to run out of research project suggestions. He always had a keen lookout for the state of the art in the AI industry and was up-to-date with the trends and research work. This is an important quality that most active students interested in AI look for — a person with a good background in AI who can constantly suggest interesting problems to work on.
Much of my time in the month of December were spent in reading recent research papers (some of which were released barely 1–2 weeks before!) and to try and implement these research papers bottom-up, from scratch, using frameworks like Theano and Caffe.
January marked the beginning of projects where I invested a little more time in learning the basics of Deep Learning-based Computer Vision techniques such as Convolutional Neural Networks, and getting hands-on with the company’s business problems. This transition seemed pretty seamless with the help of Vivek who was already quite experienced in the field of CV. The major problem we tried to solve in this period was to automatically recommend products identified from a real-world image. This is also where I learned the difference between algorithms presented in research papers that work just great in standard datasets and why it takes a lot more effort to build systems that have to work well in real world conditions. I was there when the project was just started and we had a half-working prototype. I'm sure the team here will take this project to production after a few months.
My experience at Artifacia is something that I definitely cherish, and I’m confident that Artifacia will grow to be a great technology company driven by innovation at its heart. Good luck, guys and thanks again for having given me this wonderful opportunity which I'm sure will help me a lot as I continue to pursue my career in machine learning.