In our commitment to recruit and support emerging talent in the fields of machine learning and computer science, Loyal provides personal and professional opportunities to graduate students from some of the top technological universities in the country. Through our chatbot platform, Guide, interns have access and opportunities to gain hands-on experience with machine learning, artificial intelligence technology, and natural language processing (NLP).
Souvik Bagchi, a Computer Science graduate student from Northwestern University, spent his summer as a machine learning software engineer intern with Loyal. Over the last few months, Souvik focused on developing and improving the model that Guide uses to recognize utterances and deliver answers to healthcare consumers.
“Souvik contributed to our AI Labs team immediately. There is so much research going on right now in artificial intelligence that it can be overwhelming to read papers and then act on them” says Abhi Sharma, Loyal’s VP of Product. “His work helped push our R&D forward, specifically our ML pipelines and model architectures. It’s a testament to him, our industry partner Northwestern University, and our ongoing relationship with Northwestern Medicine.”
We wanted to take the time to appreciate all of Souvik’s hard work and showcase some of the amazing projects he completed during his time with Loyal, as well as his future plans once he returns to Northwestern and afterward.
What led you to choose Loyal as the place you’d spend your summer with?
I wanted to work with machine learning, NLP, and developing intelligent systems. I had previous experience working with NLP, so Loyal was the perfect match to strengthen and add to my NLP skills. Loyal’s chatbot was a great opportunity to get hands-on experience in all of those areas. NLP is such an interesting field that I wanted to explore and understand better.
What were you working on over the summer?
I worked on a few projects, but the bulk of my work focused on 4 areas:
- Classifying intents – Improving the model to recognize and respond correctly to intents.
- Machine learning pipeline and word embedding – Making it easier to train the AI-model and deploy it. Also making it easier to retrain the model for constant learning.
- Data processing – Ensuring the data the model is being trained on is accurate to how real people communicate.
- Offline topic modeling – Trying to make sense of utterances that did not fall under a specific category and the bot labeled as “null”, or unknown.
What was one of the most memorable projects you worked on this summer?
The most interesting (and fun) project I got to work on involved conceptualizing the machine learning pipeline. Basically, I worked on the chatbot’s model to make it more accurate with its responses, which was a challenge, but I enjoyed the work and the ability to improve a system and product that helps healthcare consumers.
How was working with Loyal’s Product and AI-Labs team?
Loyal and the Product team were extremely supportive and collaborative. I always felt like Abhi (VP of Product) and Matt (Machine Learning Engineer) really listened and valued my ideas, and trusted me to implement and test those ideas. Overall, everyone was supportive and it was great working with a team of data scientists, engineers, and analysts from a variety of different backgrounds and experiences.
What did you like most about living in Atlanta for the summer?
The trees! It reminded me a lot of Evanston, and I loved being able to ride my bike around the city and enjoy how green Atlanta is. Atlanta has a lot of hills, so riding was tricky and fortunately, I only fell off my bike twice, so that was good. The people in Atlanta are extremely friendly and warm too, so I felt very welcomed.
What are your plans after graduation?
I’d like to work on building intelligent systems – especially in an area where it involves deploying those systems to the cloud.
