Career Development Tips from a Senior Data Scientist at Amazon

Data science as a discipline – and specific skills in machine learning, analytics, and training algorithms – are in high demand.

It’s a field that has exploded in popularity over the past decade and is expected to create 11.5 million new jobs in the United States alone by 2026.

So what’s it like to work as a data scientist, and what do you need to know if you’re considering starting your career there (or transitioning there later in life)?

I asked Naveed Ahmed Janvekar, a senior data scientist from Seattle who works in Amazon’s fraud and abuse prevention team, to share his professional journey.

Check out his story and the advice he has for those looking to pursue a career in data science.

A Spark: Using Machine Learning to Solve Real-World Problems

What led you to a career in data science?

Naveed Janvekar: My interest in machine learning grew while working for Fidelity Investments as a software developer.

I had colleagues who worked as analysts with data to identify trends, which made me curious to explore this area. So I started analyzing my personal financial transactions to generate trends and insights.

This has led to spending more time researching machine learning and knowing how to harness it to model repeating patterns to predict future outcomes and use it to our advantage to solve critical problems at scale.

In order to acquire a better expertise in this field, I decided to pursue my Master in Information Sciences with a specialization in Machine Learning and Analytics.

After graduation, I worked in various US-based companies in different analytical roles such as Analyst at Nanigans (a Boston-based AdTech startup), Business Intelligence Developer at KPMG and Senior Data Scientist at Amazon.

The role of AI in data security

What role does machine learning play in your work as a Sr. Data Scientist at Amazon?

Naveed Janvekar: Machine learning and data science play a vital role in my work at Amazon.

In the Abuse Prevention team, we use various classification algorithms and deep learning algorithms to detect fraud and abuse on the platform.

Machine learning helps achieve scalability and high-accuracy detection over traditional rule-based and/or heuristic-based abuse detection.

As abuse behaviors become complex over time, machine learning helps us meet this challenge since we are constantly retraining models with the latest abuse behaviors/patterns.

I filed patents for inventions related to the detection of emerging abuses on the platform using machine learning.

Communicate data-driven insights

What unexpected skill or experience do you think has helped you as a data science professional?

Naveed Janvekar: The ability to gain domain expertise and being able to effectively and simplistically communicate information to business stakeholders has helped me the most as a data science professional.

When I started my data science journey, I put a lot more emphasis on the technical details than on being an effective storyteller.

But over the past few years, I’ve come to realize that being able to communicate stories and insights from data science or machine learning is as important as implementing machine learning strategies.

Working with algorithms to create change

How should companies adapt their approach in this space in the future?

Naveed Janvekar: In the past, fraud prevention was traditionally done using business heuristics.

If you have observed that a certain pattern appears frequently over time, you can set a business rule to report the same pattern in the future.

However, this is a short term solution. It does not track changing fraud patterns.

This is where machine learning and AI come in and have changed the landscape.

Now, models are trained using historical data on multiple fraud behaviors, which makes these models robust and helps algorithms learn complex behaviors, which is much harder for humans to do.

Companies have started using machine learning in fraud detection. They now need to focus on things like automated model retraining to capture the latest fraud behavior and make the models highly accurate.

This automates actions following model output, rather than having human auditors needed to assess suspicious entities that are flagged after the fact.

Working with data and algorithms can be difficult

But what makes it exciting and fun?

Naveed Janvekar: I enjoyed engineering features from data, which brings out my creative side.

Based on domain expertise, data scientists can leverage data in a variety of ways to answer business stakeholder questions, perform exploratory data analysis, find correlations between variables, and conduct engineering. features for better model performance.

When it comes to algorithms, I’ve always experimented with training different types on training datasets, performing evaluations, and in-depth analysis of why certain algorithms work better than others.

It helps me better understand these algorithms and the situations where they work – and where they don’t.

All of this makes the job fun and exciting for me.

Be part of the data science community

What useful advice would you like to share with data science beginners who are interested in its applications in marketing and commerce and who may wish to improve their skills in this field?

Naveed Janvekar: A useful suggestion would be to participate in research and inventions in the field of machine learning and data science.

Be part of working groups trying to solve problems in your area of ​​interest using machine learning.

Contribute to their research, get feedback from your peers, publish articles and file patents.

Through these mechanisms, you actively contribute to the scientific community, constantly learn from your peers and improve your skills.

It’s also a good idea to have a data science mentor.

Follow SEO trends

How does a data scientist stay up-to-date and informed in the field of SEO?

Naveed Janvekar: In SEO, machine learning helps understand queries, voice search, and personalization.

Data scientists can explore the application of various cutting-edge algorithms for SEO use cases to measure the effectiveness of new algorithms.

This will allow data scientists to keep up to date with the latest industry trends, as well as update the machine learning stack in SEO-related businesses.

There are various journals and conferences, such as the IEEE International Conference, on machine learning and applications to help you learn more about the latest trends in machine learning.

It’s not directly related to SEO, but will help you understand the technological advances that will disrupt your space next.

More resources:


Feature image: Courtesy of Naveed Janvekar

Sean N. Ayres