Can you share your story about becoming a Data Science evangelist?
I graduated in 2009 with a degree in electrical engineering and joined a power distribution industry in India called Tata Power. My work mostly consisted of crunching electronic meters and network data and predicting power outages. It was all about the analysis of humongous amounts of data. However, all the data analyses were done using proprietary software.
Then, I came across the R, and I realized how the same work could be done much less time and more efficiently. During that time, I came across a MOOC called “Analytic Edge” on the EdX platform. They had real life analytics examples from sports to healthcare, and I realized that the skills that I had developed were not just for the power industry but could also be used in other domains. During this time, I also became a mother, and so I was not thinking about taking Data Science as a full-time career, just a hobby side project. I just started learning more about the tools and applications and machine learning in particular. However, I saw the potential in it, and after working for six years in the power distribution domain, I quit my job and devoted myself entirely to Data Science. That happened in 2016, and after freelancing for three years after that, I joined H20 as a Data Science Evangelist.
It is a very good life lesson for anyone quitting for work-life balance and struggling to come back. How did you find the time to devote yourself for Data science? What was the driving force?
When I left my job, I had many questions. There is always a notion, especially in core sectors, that if you take a break, you won’t get a job, and because of this stigma, people hesitate to take the plunge. My work-life balance was in tatters. I had to commute a lot in my previous job, almost 50 km a day, a challenge. On the other hand, a few of my friends in the IT field used to work from home. I thought to myself why I couldn’t switch to such a domain to maintain a work-life balance and work more efficiently. This was one of the driving forces in my mind back then to change my career.
I used to write a lot during my school years, but that suddenly died when I joined the corporate world. I always heard my teachers saying that communication skills mattered in our jobs but honestly, but I did not use them in my previous job. However, with a career in Data Science, everything came back full circle. This is something that I’d like to stress upon. Data Science is a profession which requires multiple skill sets as well offers many avenues. You can write, talk, give seminars and webinars. By marrying the knowledge of programming and data analytical skills with my writing skill years, I saw a real passion in a career in data science. I thought this was my calling.
When you are passionate about something and enjoy doing it, then age and timings don’t matter.
What are the challenges of being a Data Scientist?
Many people don’t understand that a data scientist’s work doesn’t end once you create a model. Ultimately, it has to be converted into business metrics to understand how it affects investment return. For this specific field, you cannot work in silos. It takes some time and collaboration to gain experience in this field. Another important consideration is to keep yourself updated with the latest trends in the industry. State of the art keeps changing every month or rather a week. You have to give yourself a sufficient amount of time to master the skills. You have to learn in this field continuously, and for that, you need to be self-motivated.
Self-motivation, perseverance is an essential aspect to sustain in your journey through Data Science.
What is your advice for people who wants to change their role or even industry, like people working in banking and electrical sector want to get into data science?
My advice for people switching domains is that you should use the domain expertise that you have gathered over the years. For instance, a person’s domain knowledge from the banking industry will be an excellent asset for the data science team. So domain experience is something you should take with you and not throw away.
Many people create a data science portfolio consisting of projects that don’t relate to their background. If you are from a healthcare background, focus on healthcare case studies. If you have experience in the BFSI sector, create projects dealing with insurance and finance. Don’t be disheartened that you are starting late in data science. I also entered this field after seven years in the power industry. I made sure that the skills I learned in my previous job were used as an add-on to my data science career.
During your Data Science journey, what challenged you the most and how did you overcome those challenges?
It is the imposter syndrome that strikes you again and again. You always think that you did not graduate as a computer science engineer, or you do not have a statistical background, and so on. Every time this thought strikes, you start to doubt yourself. This is very normal, and it happens in every industry. For women, it happens a lot. We tend to underestimate ourselves much more, and personally, It took some time for me, too, to come out of it.
If you think you are weak at a concept, then work upon it. If you have a problem, don’t believe that you are the only one with that problem. Just reach out to the community. You will notice that there is a lot of support out there.
Be open and dont be afraid to ask for help.
So far, we talked about the challenges of a Data Scientist. How about enterprises adopting Data Science? What are the challenges facing enterprises venturing into Data Science?
There are a lot of challenges when it comes to creating an AI ecosystem in a company. All these challenges can be summed under three main headings, i.e., Talent, Time, and Trust. Putting together a team of talented individuals is one of the most critical concerns for many businesses today. Another crucial element is time. It is essential to see how fast you can get business results by implementing an AI strategy. Then there is the issue of Trust, i.e., trust in your Machine Learning models and your ability to explain your models’ results to regulators and stakeholders.
I think AutoML is a promising field that provides an answer to a lot of problems. AutoML is fundamentally changing the face of ML-based solutions today by enabling people from diverse backgrounds to use machine learning models to address complex scenarios. Many AutoML tools are out there and make it more straightforward for people and companies to use. AutoML adoption is on the rise, and it’s a way to democratize AI in the real sense. That is the future that I am looking forward to.
As the Data Science adoption grows, what are the challenges enterprises should be aware of?
There are many challenges that the enterprises are facing right now; some of the major ones would be getting access to the right kind of data and getting data science outputs into production. I think deployment is one of the biggest challenges affecting the data science industry.
There are a lot of factors affecting models pushed to production. Getting data science outputs into production, where they can impact a business, isn’t always straightforward. A lot of time is spent getting data ready (loading and cleansing) before using it to develop models and visualizations. Once their models are ready for production, they contend with numerous environments, dependencies, and even skill gaps before models see the light of day.
Concerns about bias and privacy are on data professionals’ minds today, with a lot of companies feeling that these are the biggest problems to tackle in the AI/ML arena today.
That is a big challenge the whole industry is facing today!
With so many vendors offering Machine Learning tools, enterprises are often confused with all the offerings out there and which one to use? How to handle such situations?
Deciding which technologies to use is an important consideration that can profoundly impact your business. If companies start on the AI journey, they may need to determine if they want to go with open-source software or closed source or maybe both. Many existing pioneers in machine learning and AI are regularly open-sourcing their technologies, which could act as a good starting point for others. Once these new AI players have matured, they might need a vendor to support them. One can also choose a vendor who is itself the maintainer of the package; for instance, for open source H2O-3, H2O.ai is both the maintainer and vendor.
Another point of differentiation could be how fast a company wants to get started with the technology. If you start fresh and do not have an existing dev-ops system in place, it will be easier to get started on the cloud. This will eliminate the need to procure hardware, set up the software, security, infrastructure, and maintenance issues. However, if you already have a decent DevOps infrastructure in place, the on-premise option can optimize costs. Many companies prefer the hybrid model, too, whether they switch between cloud and on-prem, as per their requirements, which is a great practice.
Current COVID situation have completely changed our lives. How did AI impact COVID? Can you share some of your thoughts?
COVID impacted the AI industry in several ways, right from how people worked to conduct business. When COVID struck, many companies went remote. They dramatically increased their spending on powerful cloud-computing technologies and migrated more of their work and communications online. As consumers bought more online to avoid the new risks of shopping in stores, the e-commerce company soared, giving sellers more data on preferences and shopping habits.
On the other hand, AI technologies were also being used to fight the pandemic too. The healthcare industry is evolving rapidly, with large volumes of data and increasing challenges in cost and patient outcomes. As a result, AI solutions proved to be very promising in the current COVID situation. AI techniques helped predict and identify COVID-19 patients under intensive care who were most at risk for respiratory failure or insufficient blood flow. Identifying high-risk patients, prioritizing treatment based on patient acuity, and reducing the false alarm rate likely provide better care for patients and reduces the strain on the ICU personnel.
Machine Learning techniques are assisting healthcare professionals in making decisions faster with accurate predictions.
What does AI adoption looks like in different countries?
AI adoption is increasing, and more nations are becoming aware of the potential it has. However, a lot of barriers remain. AI early adopters in some countries are more likely to use AI to create a strong competitive advantage.
As far as India is concerned, AI adoption is primarily driven by the large global technology conglomerates, selected startups, and the GICs/GCCs (Global Capability Centers) based out of India. Some of the sectors that have adopted AI include customer service, finance, and tax, HR, IT, and cybersecurity, to name a few. Infact, India reported a 45 percent increase in AI use in 2020, the highest among all countries.
What does future for AI looks like?
Data is everywhere. With data comes the power but also the responsibility and the fear of losing privacy. AI can provide solutions and make life better, but it should be used in the right way.
Some sectors are at the start of their AI journey; others are veteran travelers. Both have a long way to go. Regardless, the impact artificial intelligence is having on our present-day lives is hard to ignore.
You are an inspiration to many with 15K+ followers and 1.8M + views in medium. Can you share your experience in writing and the people that you met along the way?
I used to make a lot of handwritten notes. One day I decided to publish them as Medium articles, one at a time. I started receiving good responses, and that gave me the confidence to continue.
When people write messages to you about how your writing helped them, that is the best compliment you can get. It motivates you to write more and even better. You also get some pretty harsh comments too, but you have to take all the positive and constructive feedback and ignore the negativity.
Congratulation on founding Women in Coding and Data Science. Can you tell me more about this foundaton and its work?
Programming is an essential skill that can open up a lot of avenues for you. Women in Coding and Data Science is a collaborative platform where women can mentor, learn, and work on projects. We have 20 mentors as of now mentoring undergraduates in colleges.
Our idea is not to have only webinars or meet-ups but also have projects to work on. For instance, we have a Blogathon for women to participate in and polish their writing skills. We all learn something from each other and, in turn, help others in need.
It was nice talking to you and knowing different aspects of your career graph. Continue inspiring us and Good luck for all your work.