We are really excited to present to you our conversation with Rutuja Udyawar, the Founder and Director of Optimum Data Analytics.
In this conversation, Rutuja deep dives into the Software Life Cycle Development of AI products, advice for individuals and teams who wants to start tech startups, in-depth pragmatic advice for women who wants to get started with AI and data science and more.
SDD: Rutuja, your extensive experience in the field of data science as a data scientist is inspiring. What was your first encounter with your field like and can you briefly describe your journey so far?
RU: I have a fascination for numbers since childhood, and hence my inclination was to go for graduation in Mathematics. During my first year of B.Sc., I realized that Statistics is more intuitive and more applicable to real life problems than pure Mathematics. This inspired me to opt for Statistics as a major in my final year of B.Sc. That was the era when ‘Data Science’ word was not coined. But as Statisticians, we were exposed to datasets of various domains like healthcare and economics and we applied various predictive models on those. My last two years of B.Sc. laid a strong foundation, which prepared me for the opportunities in ‘Data Science’ career. In final year of B.Sc., I was among the top 10 students in Mumbai University and got felicitated by the vice-chancellor of the University. It was a very proud moment indeed. After B.Sc., even though there were options of pursuing M.B.A., which was really lucrative course at that time, I listened to my inner voice and went for M.Sc. Statistics. In the second year of my M.Sc. I had internship at Hinduja Hospital in Mumbai, which was my first experience of real healthcare data. This experience was truly enriching and also helped me to land my first full time employment at TATA Motors Limited (TML).
The formative years at TML were unprecedented in various aspects. I gained tremendous experience in applying Statistics in the manufacturing field and got to see very closely how TATA empire works; I was very thrilled when I saw Mr. Ratan Tata for the first time. During this time I got married and we decided to pursue further studies from USA. Thus, I reached to Michigan State University (MSU) for M.S. in Applied Statistics. Higher education in the USA developed my truly independent thought process. Nowadays, various types of datasets, help forums are easily available on the internet. Things are more open-sourced. This was not the case in early 2000s, but still at MSU, the courses expected real life projects. During my Masters, I also worked as a Statistical Consultant at CSTAT (Center for Statistical Training and Consulting) at MSU. CSTAT gave me first experience of service industry, where I handled clients from diverse domains like history, music and meteorology. It gave me exposure to all the statistical methods from designing surveys, sample size estimation all the way to complicated forecasting models. Strong work experience at Hinduja, TML and CSTAT, led to my campus placement as a Financial Analyst in the Analytics Department of Ford Motor Credit Company (FMCC). This was the time when ‘Data Science’ term started gaining some popularity, when Statisticians and Computer Science engineers started working together and new designation called ‘Data Scientist’ got coined. At FMCC, I bagged several departmental awards and presented various papers (including one on Natural Language Processing (NLP), pretty new at that time) in Analytics conference held by Ford. I was able to manage all this along with giving birth to my baby. Of course, great credit to my supportive husband, family and kind neighbors. I realized, determination is the key to success.
Even though my green card was filed by Ford, we decided to come back to India as we always wanted to start our own venture in India in the long term. Hence, after coming back instead of getting associated with multinational companies, I decided to work with start-ups or small companies to understand their end-to-end operations. With all this strong work experience, my start-up “Optimum Data Analytics” came into reality.
SDD: You founded Optimum Data Analytics last year that focuses on delivering innovative AI solutions. What inspired you to embark on this journey?
RU: When I completed my M.Sc. from Mumbai University many people didn’t even know the word ‘Artificial Intelligence’. But, when I came back to India from USA, there were hardly anyone who had not heard the word ‘Data Science’ or ‘Artificial Intelligence’. The AI solutions that are well known today, are based on data. And as a Statistician, of course that fascinates me. So, that’s why Optimum Data Analytics (ODA) is into AI.
Inspirations come either from idols or from philosophies.
- TATA group is always my idol. During my TML period, I read books and articles about the Tata family. The autobiography of Jamsetji Tata and JRD Tata helped me understand the importance of wealth creation for the country.
- And here comes the philosophical inspiration. Working in a job, even at a senior position, still keeps us dependent on the goals of that organization. In my case, I always wanted to form my own company, even from TML period. Every entrepreneurial mind has an inclination to create her/ his own entity. But, not everyone can give justice to that inclination. Working in a self-owned company is like raising your own baby. And what else can make a mother happier.
SDD: Based on your experience, what are some unique features of product development with AI technologies? Do you have any advice for teams trying to create innovative products using AI?
RU: Unique features of product development with AI:
- As with any software product, an AI product also undergoes a Software Life Cycle Development (SLCD) which consists of backend, middle ware, front end and testing. Apart from these, as AI is based on data, an AI product has to take care of entire data pipeline, including the feedback loop which keeps updated real time. Hence, the sellable product needs to have an infrastructure which can take care of data storage and database schema. In summary, AI product development has added dimensions compared to usual software development due to which it is called as ‘Technical Debt’. Ref: https://research.google/pubs/pub43146/
- Bigger the data, heavier are the models. That’s why infrastructure requirements of AI systems are usually more demanding than other systems. If not invested appropriately, it may really be a big bottleneck, especially if the application is supposed to run real-time.
- Usually AI is synonymous with Python programming language. However, when data size grows to 100s of GB then we may need to adopt the big data stack like Hadoop, HBase, Scala and Spark.
- In usual SLCD, Proof of Concept (POC) is supposed to be precursor of project. In AI product, POC may not give a complete idea of project as there could be a lot of variation in data and in POC, you are exposed only with small sample of data having limited variation.
- AI product needs collaboration between data scientist and data engineer. While data scientist develops the model, data engineer helps to integrate the whole data pipeline and bring the model to operation.
- There are no standardized guidelines for creating test use cases for AI product till date. AI testing is heavily influenced by the Data Scientists who are essentially developers of the product. They bring their own biases, thus, affecting the quality of product.
- AI is still in a research stage. Hence, exact effort estimation is challenging due to which costing and profit calculation may go haywire. Also, due to scarcity of good data scientists, AI product development is costly affair.
Advice for teams:
- As AI is still in its nascent stage, there is still a lot of scope to come up with unique AI products. But at the same time, thorough market research before development helps a lot.
- The time between research and its implementation is far less than it was ever before. Hence, the product based on a particular algorithm could be completely outdated tomorrow. Hence, you have to be always in touch with all the new developments.
- Instead of developing the product in isolation, have domain experts around you. They are better people to register their requirements, their pain points. These requirements are actually the features that you want to incorporate into your products.
SDD: From an entrepreneurial and growth (particularly funding, talent acquisition etc.) perspective, what advice do you have for women who are interested in setting up tech startups?
RU: A tech start-up needs various resources, especially when you want to make it big and scalable. Money and talent are the most important of those. But in both of these aspects, founding a start-up and taking it to the next level is much easier than ever before.
Let’s consider first the money part:
You have to be clear whether the start-up is bootstrapped or funded. The cost is important part of any product development.
The government of India has started multiple programs like Invest India and Start-up India (‘Start-ups’ get special benefits in taxes). Women entrepreneurs get loans at lower interest rates.
Invest India (https://headstart.in/) is formed to bridge the gap between investors and innovators. Also, there are multiple venture centers funded by Government of India which incubate nascent companies and help them grow.
Apart from these, there are many angel investors and Venture Capital Investors’ who could be interested in your product. There are several start-up conferences held in cities all over India, which evaluate your ideas and team strength and fund new innovative projects. Key is to have a detailed business plan and have your sales pitch ready.
These days even MNCs are hiring small start-ups if they are into niche area.
Now, the talent acquisition part:
Brand yourself and your company on social media, especially on LinkedIn. That helps attract talent.
As a start-up, unless it is extremely well funded or backed by a big company, getting experienced people are not affordable. Hence, I believe hiring fresh talent and grooming them. I select young people, who are careful about their career, start working towards online projects and courses of their choice from the very beginning of their bachelors course. They are very passionate and have a lot of enthusiasm to work on new ideas.
Also there are plenty of freelancers and consultants who I work with. These experts are a great guidance to my young team.
SDD: There are several women in our community, who enroll in online courses to learn data science and complete them. However, once the course is over they are unsure about how to apply those skills.
RU: Apart from finishing these online courses, what else should they be doing so that they can develop skills that can be readily applied on projects?
- During the course one must keep thinking about all the use cases where those concepts can be applied. In fact, before beginning the course itself, if a proper use case can be identified and the corresponding data is selected, then all the concepts can be implemented on that data since the beginning.
- Practice is key. Even if you have inclination or not, if you keep practicing you become better. When I say practice, it is in the context of actual data analysis. Websites like kaggle, Analyticsvidya, Hackerrank provide data science challenge. Recently Google has released millions of free datasets. Participate in AI hackathons.
- Start looking for opportunities. Interactions during interviews with industry experts give better understanding of their true pain points.
- Also, many start-ups like Optimum Data Analytics are always on the hunt of talent. Most of the interns that are working with us have done one or two courses and now company offers them real life use cases on which they are building POCs. They are doing this while balancing all their college sessions, other professions or family issues. The only motivation is passion towards ‘Data Science’ field. We have worked successfully in this lockdown situation too by using various communication channels.
In summary, enrolling in a course is like getting your feet wet as that is more like getting exposure to the data science field. But unless you immerse yourself and start working on the opportunity, you will never get in depth feel of it.
SDD: Are there some of the new research papers or projects in AI that caught your interest recently, if there are can you share with us?
RU: I have installed app of arxiv.org in my mobile. That helps me to keep up with the recent research.
Here are few reference material I want to share:
- Mathematics for machine learning: https://github.com/mml-book/mml-book.github.io
- Machine Learning Testing: Survey, Landscapes and Horizon: https://arxiv.org/abs/1906.10742
- The Hundred-Page Machine Learning Book by Andriy Burkov http://themlbook.com/wiki/doku.php
- Graph Convolution for Multimodal Information Extraction from Visually Rich Documents: https://arxiv.org/pdf/1903.11279v1.pdf
- International evaluation of an AI system for breast cancer screening: https://www.nature.com/articles/s41586-019-1799-6
SDD: What inspires you every day?
RU: My ‘to do’ list!! I enjoy my work. I am associated to various organizations and various like-minded people. Each of these organizations and my colleagues keeps giving me new opportunities and challenges every other day. I accept only those opportunities that are truly meaningful to me.
As a business owner, I have to keep the balance sheet in order. But some opportunities are beyond the balance sheet. Those are for mental satisfaction. I always keep a mix of both these types of opportunities and that keeps me inspired to look forward to the next beautiful day. Tremendous support from my family, my very positive and passionate colleagues, and all those youngsters who look up to me as role model inspire me every day!!
Visit Optimum Data Analytics at : www.odaml.com
You can also reach Rutuja for business collaboration or career guidance at firstname.lastname@example.org.
Join our She Drives Data community on SHEROES (You can also download the app on iOS and Android) to connect with thousands of women data enthusiasts across the world, hear from data science experts and get updated on the latest tech news every day ❤️️