To understand this, Analytics India Magazine got in touch with Sadaf Sayyad, data scientist at Intuit, who walked us through a typical day at her work, alongside sharing interesting instances, career growth, and the impact she is adding to the team and the ecosystem.
“For a data scientist, a typical day depends on the phase of the project one is working on. But, on a high level, my day starts with checking emails and messages for any urgent tasks. Then, we have a stand-up meeting to discuss the progress of the project and blockers followed by planning my day,” said Sayyad.
Sign up for your weekly dose of what's up in emerging technology.
Further explaining, she said that the tasks include finding data sources, cleaning data, performing exploratory analysis, designing the ML-based system, which includes inputs, outputs, success metrics, building and experimenting with different ML models to optimise the target metrics, designing and conducting experiments to prove model performance in production, representing the model results and insights to business stakeholders and collaborating with machine learning engineers for the deployment of models.
“Depending on the phase we are in, my day involves one or more of these tasks. Apart from work, I carve out time for reading research papers, keeping myself updated with the latest developments, peer to peer learning through lunch and learn sessions and conferences. We also have fun at team outings, games, and even online team games in work from home setup,” said Sayyad.
Sayyad completed a master of management from the Indian Institute of Science (IISc), Bengaluru. Her elective courses focused on analytics, data science, and machine learning. Post this; she got a campus placement in Walmart, where she got the opportunity to work on projects in optimisation and machine learning. After that, she worked at LinkedIn in the data science team, where she was responsible for deep-dive analysis and experimentation of features on LinkedIn jobs pages. “This was a stint where I learnt more about key business metrics, stakeholder management, product ownership and power of data insights to drive business decisions,” she added.
Sayyad told AIM that she would have been a quantitative financial analyst if she wasn’t a data scientist. “I developed an interest in finance during an internship in a hedge fund and would have pursued it further if I hadn’t been a data scientist,” said Sayyad.
Hoops and hurdles
“The challenge and beauty of being a data scientist are that every problem you get is likely different. Therefore, the single approach most likely will not work for two problems. This makes our job very exciting as every project is a new learning opportunity,” said Sayyad.
Further elaborating, she said that on a high level, the steps or processes to follow are similar – i.e. define a problem statement and set clear expectations, ensure we have the right quality and quality of data, and set success metrics. “We build the first version model/solution as a proof of concept to ensure there is merit in pursuing a project. Then, suppose the target metrics look positive, and the cost of building and maintaining a model is worth the benefit. In that case, we go-ahead to build out a production-level model,” added Sayyad.
Overcoming data science block
Often data scientists are under pressure or overburdened with work/tasks leading to data science blocks, which could hamper their daily activities. However, Sayyad said that she overcomes this with the spirit of teamwork.
“This depends on what is the cause of the block. Sometimes the block is due to lack of data; in that case, we communicate with others to find alternative data sources, and if not, talk to business stakeholders about what is the best approach we can use and what is the best we can deliver with the available data and resources. The other block could be when one is struck at a model accuracy, which does not seem to improve even after multiple approaches,” explained Sayyad.
She said that this is when it might help to get a fresh perspective, and team knowledge helps. And talking to other data scientists about the approach one has taken and what new things could be tried can bring us back on track.
Motivation at work
“Knowing that I am working on a product that affects people’s lives in a meaningful way by powering prosperity to small businesses and customers is undoubtedly the biggest motivating factor,” said Sayyad. In addition, as a data scientist, she said that getting to solve exciting problems and learning something new every day is a great motivating factor.
Sayyad said she wants to grow her technical expertise in artificial intelligence, keep herself updated with the ongoing research and contribute and give back to the AI community.
“I want to continue making an impact in people’s lives through the power of AI. With Intuit’s strategy being an ‘AI-driven expert platform,’ I couldn’t be better paced,” said Sayyad.
Work at Intuit
“At Intuit, my role has evolved from building ML models focusing on the technical aspects and algorithms to extending this to building reusable AI-based systems focusing on improving customer experience and ease of use,” shared Sayyad.
She said that she had had multiple opportunities to work on some amazing and impactful projects at Intuit, delivering key success metrics for the company and learning and implementing state-of-the-art ML techniques that have helped her grow as a technologist.
“I am currently working on a project which will help us improve customer experience significantly as they provide resolution to the problem they raise, using the power of AI,” said Sayyad. She said that she uses computer vision (optical character recognition) and natural language processing (document classification and named entity recognition techniques) for this project.
Previously, she has also worked in multivariate anomaly detection and supervised machine learning problems.
“At Intuit, I am delighted to work with a team of highly talented people where we learn from each other every day. There is no exaggeration when I say everyone personifies the company’s value of ‘Stronger Together,” said Sayyad.
Further, she said that the leadership is also very clear about the top-level goals called ‘Big Bets’ and tech priorities, and every project is aligned to these goals, so they always have their eye on the big picture and what they are working towards.
Adding to this, she said Intuit has repeatedly been in the top three best places to work in the Great Place to Work ranking because of its employee-first and empathy-driven policies.
The AI and data science team at Intuit is currently about 500 members strong, distributed across multiple geographical locations. For example, the team in India consists of data scientists, machine learning engineers, machine learning infrastructure engineers, business analysts and programme managers.
Sayyad said that there is a lot of encouragement and opportunity for peer-to-peer learning. “We have a cadence of knowledge sharing sessions within our team and have resources available to learn what other members have worked on. Contributing in these forums and sharing valuable feedback is one of the ways we contribute to each other’s success,” she added.