Machine Learning Research Engineer Interview

AI based medical research is expediting end-to-end drug discovery and development through ML techniques that can uncover complex antibody-antigen interactions.Saheli Mitra, our next pathbreaker, works as Machine Learning Research Engineer at MAbSilico (Tours, France), an organization dedicated to the medical research through algorithm based solutions for developability of antibodies.Saheli talks toShyam Krishnamurthy from The Interview Portal about her research experiences during her masters at IIT Kharagpur that paved the way to a PhD in Soft Condensed Matter Physics aided by machine learning approaches.For students, AI technologies are unravelling biology in ways that were never possible before, thus bridging the gap between research and its applications in society !I was born and brought up in Kolkata, West Bengal. We are a family of four, my parents and my younger sister. My father worked as a clerk in an electricity company, and my mother was managing the household. We had a very simple lifestyle, we knew on two occasions of the year ( bengali new year and durga puja) we would buy clothes. We bought new pairs of shoes only when the previous ones were broken. We had television when I was 11 years old and two months prior to final exams in school, the cable connection would be cut off. My parents had always put emphasis on the importance of good education and “standing up on your own feet”, meaning being financially independent. There was no pressure to be among the top ranked students in my class, though our school had a ranking system. Lucky me, they did not hope much ! My schooling was from Ramakrishna Sarada Mission Sister Nivedita Girls’ School, in Bagbazar Kolkata. Our school is known for its high standard of discipline and teachers who did their best to teach us. During my school years, my father would often tell me that Mathematics and English are the “main” subjects to master no matter what I chose to do later. He also told me, it’s important to question everything, through logical reasoning. So, I logically decided that my love for physics was greater than my love for mathematics, and I would like to study physics. As for english, my schooling was in bengali medium, we learned grammar, to write, but not to speak english. whereas I speak english 98% of the time nowadays !What did you do for graduation/post graduation?The former school I mentioned did not have a high school at the time. So, after the board exam (Madhyamik), I studied science (Mathematics, Physics, Chemistry, Biology along with English and Bengali) from Baghbazar Multipurpose Girls’ School until the next board exam two years later. In those two years I had become even more determined to study Physics. I liked mechanics, thermodynamics, and problem solving in general. I graduated with Physics honors from Vidyasagar College, University of Calcutta. In my final year of graduation, some of my friends from college were preparing for IIT-JAM, an entrance exam to do masters from one of the IITs. I thought why not try? And it turned out to be the biggest turning point of my career. Thankfully, I ranked among top 100 and I did my masters in Physics from Indian Institute of Technology, Kharagpur.What made you choose such an offbeat, unconventional and unusual career?The reason I chose to study Physics was my genuine interest in Physics. And the person I can’t help mentioning is our professor Dr. Saugata Bhattacharyya from Vidyasagar College. He would engage us in discussions in the teachers’ room, that would range from explanations of something he taught in class, stories about other physicists, to life in general. When I had a little dilemma about leaving my house to go to study in IIT Kgp or doing MSc from somewhere in Kolkata, he simply said that it’s my time to explore the world outside of what is known, what is home. It’s an opportunity to take.It was a new life on the campus of IIT, mingling with students from all states, different cultures. For the first time, my classes were completely in English.In the second year, my two roommates were from the Chemistry and Mathematics Dept. They were among the toppers in their classes. This second girl was planning to do a PhD from abroad. That’s when it struck me, why not try?Thus, I ended up obtaining a PhD from University of Paris-Saclay in Physics. To be precise, my thesis was on a computational study of structural changes in glassy systems when they are subjected to periodic shear. By the way, that friend from Mathematics finally did her PhD in Germany.How did you plan the steps to get into the career you wanted?Or how did you make a transition to a new career? Tell us aboutyour career pathWhen I thought of applying to do a PhD in Abroad, I first spoke to Dr. Saugata. He gave me two contacts of senior graduate students from our college who had gone to Europe for their doctoral or postdoctoral studies. So I spoke to them for guidance on how to apply, what sort of things stand out in an application. During my masters, I had done a summer research internship and a thesis in the second year.For my master’s thesis I worked on pattern formation in fluids resulting from Rayleigh-Benard Magneto-convection under the supervision of Prof. Krishna Kumar who had given me his letter of recommendation for PhD applications.So, these research experiences did help while applying for PhD. And I had an INSPIRE SHE scholarship during my graduation and post-graduation years. I saw an announcement of an open doctoral position from a mailing list in IIT, I submitted my CV and got a call for interview with my future PhD supervisor, Prof. Giuseppe Foffi, University of Paris-Saclay. This position was funded by CEFIPRA (Indo-French Centre for the promotion of Advanced Reseach). As part of this India-France collaboration, my other PhD supervisor from India was Prof. Srikanth Sastry, JNCASR, Bangalore. He had a second interview with me and I spent three months at his lab before leaving for france. In both of those interviews I had to explain in detail my master’s thesis even though the topic was different from what I would do in PhD, which is working on glassy systems, soft matter physics.How did you get your first break?During my PhD, I was part of a project that involved use of machine learning approaches to study local structures in supercooled liquids.My PhD work was theoretical. Someday it might have some applications (like what happens with mathematics etc ), but I cannot think of anything right now.This work was published in Nature communications. Towards the end of my PhD, I wanted to know more about artificial intelligence and its applications. I could see it evolving as a very powerful tool in every field, with demand growing in the future. I enrolled in an online course from Coursera by Andrew Ng. In the end, I decided I will look for jobs that use AI after my PhD is over.My friend Pabitra (A former postdoc from my lab in Paris) gave his guidance during my job search, he had just got an industrial position in Paris. I was sending my CV to several job openings that I found via LinkedIn. Finally, I had my interview with my current company. It was during covid. First interview was via video-call. Afterwards, my current boss sent me a bunch of research papers on monoclonal antibodies. Then was my walk-in interview, explaining to me the project they had in mind for me, asking me how I think of approaching that problem. One week later, they offered me the job at MAbSilico.What were some of the challenges you faced? How did you address them?I don’t have a background in immunology, nor in Data Science. And this job needs knowledge of both because it involves researching and developing computational solutions to help the discovery process of monoclonal antibodies against different target antigens that cause diseases, using AI. So, as for biology, when I have doubt I ask for resources or speak to the biology team members. When it is related to AI, many resources are available including online forums. I can also discuss them with my colleagues.Where do you work now?What problems do you solve?I am working in MAbSilico, Tours. It is in a city 250 km away from Paris. The problems I work on are related to the interaction between proteins (antigen and antibody), developability of antibodies, and maturation. It involves understanding of protein structures, basic biology, data analysis, developing codes, use of machine learning, deep learning frameworks. I learn new things as needed from the internet and research papers during the development process.What’s a typical day like?On a typical day, I will probably spend some time in ongoing R&D projects and some time in applying the already developed solutions for projects with our customers. What I like about this job is the freedom to try out new ideas. We have “brainstorming” sessions on a white board with my team members. Since this is research, there is a good chance that not all ideas are going to be useful and that’s okay.How does your work benefit society?Since this work is on developing an antibody cure against diseases, I hope one day one of the projects I have worked on will be used in an approved drug.Tell us an example of a specific memorable work you did that is very close to you!Recently, I developed a binding affinity prediction code for antibody-antigen interaction. It is special to me because it’s my first actual work using AI.Your advice to students based on your experience?Don’t assume something is beyond your capabilities. Reach out to people who have been in the path you are thinking of taking, assess your chances realistically. Be flexible in career choices.Future Plans?For the moment I think I will have a career in data science, unless something different interests me more .
#J-18808-Ljbffr