The agricultural sector is reaping the benefits of data science based approaches through improved yields and reduced costs for small and marginal farmers.

Ajay Natteri, our next pathbreaker, Data Scientist, develops algorithms for solving various use cases in the agriculture and aerospace domains.

Ajay talks to Shyam Krishnamurthy from The Interview Portal about his first project at Mahindra and Mahindra Limited [Automotive, Farm Equipment and Agri Business] where he worked on algorithms to determine the usage of tractors, with data from telematics devices fitted on these farm vehicles.

For students, Data Science opportunities are abundant across all sectors today. Focus on the bigger picture – Who are your customers, what are your objectives, what are you trying to solve?

Ajay, what can you tell us about yourself?

I was born and raised in Chennai. I have lived and worked in Chennai all my life except for 2 years in between (2014-2016) when I lived in the US while pursuing my Master’s degree. 

During my school days, I knew of very few career options. With my dad being a Mechanical Engineer, I too opted for Mechanical Engineering once I finished high school.

While in college, I spent a lot of time in extra curricular activities – everything from volunteering with NSS & Rotaract and organizing events in cultural and technical fests. Thanks to these extra curricular activities, my soft skills improved considerably. I also ensured that my grades were reasonable for placements/higher education. My final year projects were in the Thermal Sciences stream. One project was in designing and fabricating a heating chamber and other on Engines. 

Through my interactions with college seniors, I learnt that students recruited as Engineers also perform other roles in companies such as supply chain, project management, etc. I felt taking management courses will help me in my career and applied for Engineering – Management programs in the US for my Master’s. 

What did you do for graduation/post graduation?

I did my bachelors degree in Mechanical Engineering from Guindy College of Engineering, Chennai.

I hold a Master’s degree in Systems Engineering and Management from UT Dallas.

UT Dallas offered an excellent degree in Industrial Systems Engineering with Management. As a part of the curriculum,  students must choose 4 Engineering and 4 Management courses. This was an ideal course for me plus the scholarship at entrance made my decision to choose UTD an easy one. 

How did you end up in such an offbeat, unconventional and unique career?

My first course at UT Dallas is a key reason for my career choice. The course was on Mathematical Optimization. The weekly assignments used to be quite difficult and I would spend an entire week with my friends to solve them. I started to enjoy the course and eventually topped the class. I subsequently took advanced courses in probability and statistics.  

When I started as an Automotive analyst with Frost & Sullivan (my first company), there was a lot of talk about Machine Learning. Naturally, I became quite curious and enrolled in the Machine Learning course offered by Prof. Andrew NG on Coursera. To my surprise, I already knew most of the concepts taught in the course. When I realized that opportunities for Data Scientists are abundant in India, I decided to quit my analyst role and move into the data science domain. 

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 about your career path

I had done two internships – one with Frost & Sullivan (F&S) while pursuing my Bachelor’s degree and another during my Master’s at Federal Mogul Powertrain in Mechanical Design. 

Since I had some working knowledge in the automotive domain, I got a call from Frost & Sullivan (F&S) for a full time offer within the mobility division. 

Looking back, Frost & Sullivan was a great first job. I was a part of a group called Business Strategy and Innovation within the mobility division.  My group tracked tech giants (Google, Apple, etc) and start-ups that were disrupting the automotive space. 

Just to give you a flavor – we worked on a report that analyzed the product portfolio of Google. We showed how Google is becoming a lifestyle brand with its products being used by consumers throughout the day – whether you are at home (E.g. Nest) or at work (E.g. Google Productivity tools, cloud etc) and on the go (Android Auto and now Waymo). We dived deep into aspects such as how Google can leverage data they collect when you are at home, on the go and at work to provide personalized services, location based ads. An example could be monitoring the time you leave to work in the morning and sending a waymo autonomous vehicle to your doorstep at the right time every morning without the need for booking a ride.  

There are several benefits to working at a market research firm: 

  1. You gain knowledge on the latest trends in the domain.
  2. You gain the ability to assimilate information on a topic and present it in a cogent and concise manner. 
  3. Learn the art of making good presentations depending on the target audience (a skill that helped me a lot during meetings with senior management at Mahindra).
  4. Particularly at Frost, even as an Analyst, one gets an opportunity to speak with CXO level executives across various companies – a rare experience for a fresher.  

Following my stint with F&S, I worked at ElektroThermalKinetics (ETK) as a Site-Engineer. At ETK it was to overlook erection commissioning at a vendor’s site.

Once I decided to move into Data Science, I looked at job openings in various companies to understand the skill they look for in a Data Scientist. Although I knew the math concepts to an extent, I lacked the required programming skills. 

I spent an entire year learning to program from youtube videos and practicing on whatever I learnt on dummy datasets. For advanced concepts in AI, I audited courses on NPTEL,Coursera. 

Through referral, I had the opportunity to interview with Mahindra’s Precision Farming Division – my break into the Data Science space. 

I spent 3 amazing years developing algorithms for solving various use-cases in the agri- domain. 

My first project was solving one of the key challenges that tractor owners face in India. Tractor owners generally employ a driver to till their field and are paid hourly. Thus, drivers have a tendency to prolong the operation to make more money. This affects the input cost for the farmer. We developed a solution with a telematics device at its core. By analyzing the data from the vehicle we are now able to track their tractor in realtime and also get an update on the area covered in near-real time.

Another project that I worked on was aimed at reducing the dependency on farm labor for pesticide spraying. Farm laborers are reducing by the day and their wages are increasing. A solution for this was targeted spraying of pesticides on diseased plants using robots. Robots with cameras capture images on the go and an AI algorithm identifies disease plants. The sprayer then triggers and sprays pesticide on the target plant. This serves 2 purposes – one, Reduces dependency on farm labor and two utilizes little fertilizer as the spraying is done on diseased crops alone.

Several legacy companies and startups globally are investing millions into developing tech-based solutions for the agri-domain. The ultimate vision is to use technology to fully manage a field that will in turn enable higher yields and profits for farmers. 

I am currently a Data Scientist at Rolls-Royce working on use cases related to aircraft engines. 

How did you get your first break? 

My first job at F&S was through a job posting site. 

My subsequent roles in Data Science roles at Mahindra and RR (Rolls-Royce) were through referrals.

What were some of the challenges you faced? How did you address them?

Challenge 1: Lack of Domain knowledge – Developing models with little to no understanding of the domain and its constraints adds no value to the customer. In my case, I spent ample time speaking with subject matter experts and end-customers about the problem, its context and constraints before proposing any solution. 

Challenge 2: Landing a data science role – When I reached out to companies that provide analytics as a service, my resume was never considered as they would expect prior experience with a whole bunch of tools. OEMs (both Mahindra and Rolls-Royce in my case) on the other hand welcomed candidates with diverse backgrounds and focused on problem solving skills. 

Challenge 3: Interacting with Multidisciplinary Teams – Any problem that we work on requires interaction with experts across multiple domains. For example, I have worked on projects wherein I had to work closely with electronics and robotics experts. Before any discussion, I would generally do some light reading to follow along.  

Where do you work now? 

I currently work with Rolls-Royce. Since I just joined, let me talk about my previous role with Mahindra.

What problems did you solve in your earlier role at Mahindra?

At Mahindra, we tried solving challenges faced by farmers with technology. 

What skills are needed in your role? How did you acquire the skills?

Technical skills and sector knowledge are both required. In my case, an example of technical skills are Python programming, statistics and knowledge of the Agri sector.  

Technical skills can be acquired from online courses, reliable youtube channels. One can also reach out to their contacts in the industry for guidance. 

For sector knowledge, senior colleagues in the organization generally offer ample training for new joinees.  

What’s a typical day like?

60% of the time is spent on model development. About 20% is spent on field testing. Remaining 20% is split between new requirement gathering and/or supplier management and/or meetings.

What is it you love about this job? 

I work on projects that directly impact farmer’s income.

How does your work benefit society?

Data Science opportunities are abundant across all sectors today – from aerospace to agriculture. In agriculture for example, data driven approaches actually improve yield and reduce input cost for the small and marginal farmers. In the aerospace domain, analytics based work has resulted in improved safety, higher efficiency and lower fuel consumption. In general, the use-cases are very interesting and challenging irrespective of the sector. 

Tell us an example of a specific memorable work you did that is very close to you!

At Mahindra, my first project was to develop an algorithm to determine the usage of tractors with data from telematics devices fitted on vehicles. Farmers use the application to determine the use/abuse of their vehicles when leased. Additionally, they get to charge users based on the usage – enabling usage based pricing. This entire system that goes behind this project has been patented and currently serves farmers across India.

Your advice to students based on your experience?

  • Improve communication skills: Do not neglect communication classes in college/ school. 
  • Network and learn from practitioners in the industry. 
  • Spend time to upgrade your technical skills. You need NOT spend a lot of money on online courses. There are professors from reputed institutes who put out free content on their personal website/ youtube. 
  • If you wish to enter the Data Science space, do not blindly use libraries. Instead try understanding the math behind  models used. 
  • Always have a holistic view of the project you work on. Who are your customers, what are your objectives, what are you trying to solve, the timeline and so on. 
  • Be open to new ideas and opinions from your peers, collaborate and learn from others in your team.  

Future Plans?

I plan to stick with Data Science for the foreseeable future.