Data scientists cannot grow professionally if they do not have the support of the organisation they are working in. Mere mastering a few tools and techniques will enable them to make strides in their data science journey. Apart from the data science domain knowledge, aspirants should focus on company culture and the facilities that the firm provides for them to flourish. However, data scientists often randomly apply for jobs without evaluating the company, which they later repent. This not only causes hindrance in their career but also negatively impacts the organisation.
Since data scientists ignore such thought processes, Analytics India Magazine, every week, gets in touch with prominent data scientists for the weekly column My Journey In Data Science, where the data science leaders share their journey and also advise aspiring data scientists to guide them in the right direction for their growth.
For this week, we got in touch with Pronojit Saha, an advanced analytics practice lead at Abzooba. He has over 12 years of professional experience in various verticals like retail, healthcare and Industry 4.0. Pronojit shared his exciting journey in data science with AIM and spoke about the importance of working for the organisation that facilitates a robust data science platform for aspirants.
The Onset
“I strongly believe that proper education is the foundation for an individual’s growth and by an extension of that, the cornerstone of society. My family has always been a strong advocate of holistic education for becoming successful in life and thus emphasised, since my childhood, on the role of academics as well,” begins Pronojit. “I’d developed a keen interest in mathematics, computer science and history at school, which shaped my career so far. I had a flair for programming from a young age and went on to achieve full marks in computer science in Class 12 boards, receiving a distinction. This strong foundation helped me to get into one of the premier (and oldest) technology institutes in India — Indian Institute of Engineering Science and Technology, Shibpur from where I graduated in 2006 with a degree in IT and then went on to complete my MBA from IIT, Madras, in 2008.”
Love For Data Science
Pronojit had a penchant for mathematics and programming from his school days. During his engineering and MBA, he continued to hone these skills by taking up courses in statistics and programming. The “aha” moment happened towards the end of his MBA in 2008 when he was exposed to the Python programming language, which he took out of curiosity and was not included in the core curriculum. He discovered the harmonious ways in which analytics and programming came together and left a lasting impression on him.
Post-MBA, Pronojit consciously honed his knowledge around data science, although there weren’t any dedicated courses at that time. To start his journey in data science, he did online courses from Andrew Ng and Berkeley. And most importantly, he applied what he kept learning on various online projects available across the Internet. Even today, he continues to upskill by voraciously reading articles and research papers. “I believe that the ability to learn continuously and be on top of ongoing research are some of the most important skills that enable data scientists to stay abreast in this dynamic industry,” says Pronojit.
Pronojit’s Professional Stints
Pronojit started his professional career in process excellence, and moved to establish his startup in the supply chain industry in 2009. At the same time, he also worked as a consultant in the data analytics domain. On gaining more success as a consultant, he chose to pivot his career in data science and took up a couple of positions as a senior data scientist and finally leading to his current role of Advanced Analytics Practice Lead at Abzooba.
He has worked in the retail, healthcare and Industry 4.0 domains in his 12 years of professional career. Time series analytics is his speciality, and he has applied this along with other AI lifecycle methodologies for use cases like price optimisation, readmission prediction, predictive maintenance, churn modelling, among others.
Data Science Journey At Abzooba
It has been a very exciting and fast-paced journey for Pronojit at Abzooba in the last couple of years, but then that’s Abzooba’s DNA. Starting as a senior data scientist, he developed a price optimisation product for one of the largest supermarket chains in the US. The solution recently won the best AI-based solution in retail at Cypher, India’s largest AI conference. Since then he has gradually transitioned into his current role — Advanced Analytics Practice Lead — wherein he has also been entrusted with growing the practice within Abzooba by nurturing fresh talent, building thought leadership and enabling scalable processes.
Unique Value Proposition Of Abzooba
Talking about the edge that Abzooba provides, Pronojit said the major differentiating factor for Abzooba is its people. It consists of highly excited and motivated people who are eager to make a difference in the Analytics landscape.
Further, “The projects we do are some of the unique use-cases in the industry, and we implement the latest analytical tools in deep learning to solve these use-cases.”, says Pronojit.
Besides, Pronojit also stressed on the modus operandi of developing analytics solutions at Abzooba. “We have our in-house platform— Xpresso.ai — that facilitates the end-to-end lifecycle of AI projects. Xpresso.ai is an Integrated Development, Deployment and Management Environment (IDDME), that eases the job of developing, deploying and monitoring software projects (especially Data Science projects) to a high-availability environment, by providing development teams with tools and automated processes that encapsulate industry-standard best practices. It relieves development teams from concerns and processes of development operations (DevOps) and deployment, leaving them free to focus on solving business problems of customers.
Apart from facilitating a faster turnaround time in projects, this also fast-tracks the learning curve of a junior/mid level associate as the platform provides plethora of tools and gives ample exposure to the entire AI lifecycle, an aspect which only a handful of organisations worldwide can boast of,” explains Pronojit.
Abzooba is also ahead of the industry curve in identifying the need for Full Stack Data Scientists and has been training and certifying its associates on the same for a few years now. “Full Stack Data Scientists are individuals who handle various aspects of an AI lifecycle that has been till now generally touched upon by a business analyst, machine learning engineer and data engineer in their separate capabilities. If Data Scientist is the sexiest job of the 21st century, then Full Stack Data Scientist is definitely its future avatar”, says Pronojit.
A Typical Day At Abzooba
On a typical day, Pronojit manages a wide range of responsibilities. As project deliverables are his priority, he starts the day by devising plans for the project along with his teams. They take up the long term objectives first to ensure the team is heading towards the right direction and then map them to the immediate deliveries for himself and the teams. After that, he tries to identify and understand blockers for the teams and plans towards eradicating those. Blockers can be procedural, operational or technical.
Pronojit then focuses on activities to grow the Advanced Analytics Practice at Abzooba. These involve tasks related to recruitment, pre-sales, technology evangelisation, training, and more.
Finally, he moves on to the urgent project tasks, which are assigned to him. Pronojit usually takes these up as the last items on his list as these must be completed on that day itself; this works as a productivity hack for him. Amid all this, he tries to help his fellow associates across the organisation with their queries and ad-hoc requirements — as is common with Abzooba’s culture of collaboration.
Managing Different Technical Challenges In A Project
As popularly believed, data science is a team sport, and it’s the team, which is the first line of support for any technical challenge. “We have three main practices at Abzooba: advanced analytics, big data, and data engineering. These teams are dependent on each other to solve any technical challenges in each other’s domain. For example, in one of the advanced analytics projects, we needed to migrate traditional data science solutions to big data and cloud as a final step of the project. This presented specific challenges that were ably resolved by inputs from big data practice from time to time,” says Pronojit.
For challenges within the practice, the team has two avenues. The first and foremost are handpicked online resources that help them to quickly come to a resolution. Then, they have peers to help each other out, and this is imbibed in Abzooba’s culture. People are open to queries from others. They have streamlined processes via internal email groups/channels, wherein the entire practice shares knowledge and helps each other out for answering any technical queries.
Motivating Juniors In AI And Data Science At Abzooba
Pronojit said Abzooba as an organisation is very process-oriented; the company has a thorough training and upskilling curriculum for associates on a wide gamut of subjects covering the entire life cycle of a data science project. From time to time, he also engages with associates to understand their progress in terms of learning/training while suggesting ways and workarounds in case they stumble on roadblocks. Abzooba also has a structure in place wherein it has subject matter mentors available to associates to help them on their journey. The company also believes that it is important to make associates understand the business impact of the work the team does so that they can appreciate their contribution even better and stay motivated to work with a deeper sense of ownership.
Most Successful Project As A Data Scientist
Of the many successful projects Pronojit has been involved in, he shed light upon a decentralised data-driven and scalable price recommendation product, which provides three major functionalities:
- Recommends price for various SKUs daily across various stores, while giving full autonomy of setting the price to store managers. The product can handle 100,000 SKUs across more than 100s of stores and multiple departments daily.
- Product insights are provided through an interactive UI, over and above the price recommendation that enables store managers to drill down products that require attention and can be made more profitable.
- Dashboards are made available to highlight the projected impact to store KPIs based on the recommended price.
The major differentiation of the product is the ability to work at scale. The solution can handle 100,000s of SKUs on a daily basis for more than 100s of stores. Besides, it incorporates promotion analytics and time series models using deep learning into the pricing model, thereby making it more robust to seasonal variations.
This project was awarded the Best AI in Retail at Cypher, the largest AI conference in India.
Advice To Aspiring Data Science Professionals
In an ever-changing landscape, Ponojit stresses the importance of staying abreast with the latest methodologies and technologies. One of the essential criteria that he looks for in a new hire is how quickly an applicant can learn something new and if he/she has applied what he/she has learnt.
“Apart from this, one should always look for the business case in a problem and back-calculate on your model results to verify if it makes business sense. Above all, one should have a bias for action — be it in data science or in any walk of life — to realise one’s dream,” concludes Pronojit.
