In our weekly column A Day In The Life Of, we are trying to step into the shoes of awesome techies from various organisations and sectors who are working in emerging tech areas like big data, data analytics, artificial intelligence, machine learning and the internet of things, among others.
Sign up for your weekly dose of what's up in emerging technology.
This week Analytics India Magazine got in touch with Mumbai-based Bhaskar Dhariyal, a Machine Learning Engineer at ION Energy.
Dhariyal is a health nut and a bibliophile who likes to start his day at around 10.30 am, after eating a nutritious breakfast. “I also set aside some time for a workout and a quick reading session every morning, before leaving to work. Mornings are the time I usually feel the most productive,” he says enthusiastically. Once a week, he also participates in the meditation sessions that his company organises.
When asked about his work, Dhariyal says, “My work consists of reading research papers to implement them, holding discussions with my boss around the problems we are trying to solve. The work could be identifying the best approach to solving problems or programming, which includes data preprocessing to modelling. Usually, the data is collected is through sensors, and there is a high probability of error. Also, usually the data is not in a proper format, so data preprocessing consumes a lot of my time.”
Currently, Dhariyal is working on predicting the State of Health of batteries. “It is difficult to estimate battery health due to several internal and external factors. Internal factors may include the cell chemistry, SEI formation, etc while the external factors could be how the battery was used, the temperature it was exposed to, the charging rate, discharging rate, etc.”
“Moreover, the health of the battery reduces over time even if you leave it in a standalone form. So at the end of the day, there are various factors that need to be considered, before diving into the modelling phase,” he adds.
Dhariyal says that they recently completed their study on publicly available datasets such as NASA battery ageing data, where they created two types of models — one physics-based, and the other machine learning-based. “It was a big win for the team since we were able to produce better results in comparison to a recently published journal paper with very simple NN architecture and infusing a good amount of domain knowledge as the model’s features.”
He adds that he is working along with his team to launch their next-gen battery intelligence platform – Edison Analytics, that essentially uses battery data, leverages AI, data science and digital twin technology to predict and improve the life of lithium-ion batteries.
When asked about his favourite toolkit for an emerging tech professional, Dhariyal says, I have always used Python as my primary language for ML projects. Talking about IDE I usually work with Jupyter Notebooks and VS code simultaneously.”
Dhariyal says that working at a vibrant startup like ION Energy is unconventional and exciting. “The best part is, I’m always given the opportunity to experiment with my models, and my peers are open to listening to and implementing my ideas,” he says.
“ I’m constantly learning and always eager to learn new approaches in the field. There is always an opportunity to contribute in a different manner,” he adds.
Dhariyal says that his five-year-plan includes teaching at a personal capacity where he’s given a challenge from analytics/ computer vision/ NLP and is able to successfully overcome it. His short-term goal, for now, is to see an actual deployment of his algorithm in li-ion batteries used in EVs and ESS.