Sumedha Rai serves as a Senior Data Scientist at a micro-investments firm, where she spearheads Artificial Intelligence (AI) advancements for fraud detection and directs Natural Language Processing (NLP) initiatives. She is also an AI researcher within the healthcare domain, affiliated with the Center for Data Science at New York University and its medical school, Langone Health. Over the last few years, Sumedha has fostered numerous interdisciplinary collaborations while working with various entities, including academic institutions, corporations, hedge funds, central banks, and medical professionals. She has written insightful articles on AI for prominent media platforms and has even been featured on television, where she engaged in an interview conducted by an AI bot in India. Sumedha actively disseminates her expertise in AI by regularly interacting with and evaluating AI startups, participating in tech panels, and serving as an enthusiastic speaker at conferences centered around AI and machine learning. I connected with Sumedha through one such conference hosted in New York, where she spoke about using Natural Language Processing to harness the power of text in AI models.
Below, Sumedha talks more about her successful efforts to innovate AI and machine learning in fintech and healthcare so that both industries are freed of longstanding issues and reach new levels of productivity.
Tell us more about the AI conference and the NLP work that you presented. What other conferences can we find you at?
It was a privilege to address the MLConf 2024 in New York City. Since its inception in 2012, it has been attracting top-tier speakers from global tech behemoths, such as Microsoft, Google, Meta, and Oracle. As a leader in Natural Language Processing at my company, I provided a comprehensive overview of crafting NLP models tailored for fintech firms. From laying the groundwork for integrating AI into fintech use cases to the intricate modeling process, I navigated the fundamentals of data science and AI. Furthermore, I elaborated on leveraging Gen AI and Large Language Models to address fintech challenges, showcasing how these techniques have significantly enhanced workflow efficiency within my organization and how fintech companies can actively embrace them to increase productivity.
Looking ahead, I am excited to delve deeper into Gen AI’s potential at the AI in Finance Summit, where I will compare it with traditional NLP methodologies. Additionally, at the upcoming Data Science Salon’s NYC AI conference hosted at S&P Global’s headquarters, I will discuss strategies for understanding and mitigating bias in AI models, ensuring ethical and fair AI implementation.
How did you get into AI, and what are the industries you currently work in?
Since earning my master’s degree in Data Science (DS) and Machine Learning (ML) from New York University, I have established myself as a researcher and data scientist across the academic and industrial landscapes.
I had the honor of collaborating with lawmakers at the Reserve Bank of India for two years, where I worked at the intersection of policy and ML. Currently, I work in fintech, contributing to the operations of a micro-investments firm while also pursuing research alongside academics and neuroscientists at NYU. I am very appreciative of the diverse experiences I’ve accumulated in these pivotal industries – together, they have shaped my professional growth and expertise.
Why are you so interested in AI and NLP?
I have found that the beauty of AI lies in its sector-agnostic nature. AI skill sets are quite transferable – it’s a testament to the versatility of this field.
NLP has proven to be a game changer within my company, as its impact resonates across both our business operations and customer experience. By facilitating swift communication using this technology, I’m able to effectively bridge the gap between our customers and our business, fostering a deeper understanding and connection between the two. Through NLP-powered tools, we can promptly address customer concerns, gauge overall sentiment, and identify prevailing themes, enabling us to continuously enhance our offerings and drive innovation. The efficiency gains are immense, saving us hundreds of hours of manual labor each day. Through my talks at conferences, I aim to inspire startups and businesses to harness the power of textual data in their decision-making processes and unlock new avenues for growth.
Your work within fintech sounds exciting. What is fraud modeling, and how are your AI solutions impacting fraud detection?
Fraud prevention is a cornerstone in the operations of any financial institution. By harnessing the power of machine learning, we can combat fraudulent activities by predicting the likelihood of such events and initiating preemptive measures.
At my present company, I lead our fraud prevention endeavors by designing sophisticated anti-fraud models to detect fraudulent behavior. Using AI-models that raise red flags when suspicious activities are detected, I can intercept seemingly risky activities before they become fraudulent. That gets us ahead of fraudsters, which is obviously very important.
The impact of these efforts is substantial. For instance, one of my models that applies AI techniques to relevant data saves us over $120,000 annually in losses. This represents just a glimpse of the broader picture. Through a comprehensive suite of anti-fraud solutions, I help to safeguard thousands of dollars each year.
Moreover, the ripple effect of my models extends beyond our 4M active customers to encompass thousands within their own networks. As customers express their trust in our company, positive word-of-mouth spreads, fostering stronger long-term relationships and bolstering the company’s reputation.
How have you been able to successfully transition your skill set from finance to the healthcare sector?
At heart, AI is a foundational technology – its base incorporates mathematics, statistics, and computer science. As a result, AI is really a sector-agnostic technology, and as such, I can easily switch from industry to industry and quickly make an impact. Whether I am working in healthcare or fintech, the logical thinking and modeling process are all the same.
Consider my fraud models for a banking institution – I actually use the same logic behind a fintech fraud model that I use for catching insurance fraud in healthcare. For that reason, I can swiftly transfer my skill set – its applications are, in essence, universal.
Another example are administrative models, which can be adapted to increase staff efficiency in fintech and healthcare firms. The same is true for models that help parse and summarize legal and compliance data. The AI technology in both industries remains the same, so I can make an even bigger impact.
What is the groundbreaking work you are doing with AI in neurosurgery?
My work targets the intersection of AI, neurosurgery, and healthcare. AI is being used heavily in medicine and requires both data scientists and healthcare practitioners. For instance, while building an AI model for use in neurosurgery, I do the analysis, predictive modeling, and mathematical review of results, but a neurosurgeon, who has the domain expertise to medically interpret the results, is also critical.
One of my AI projects involves predicting any misinformation that can be spread through language models in medicine and neurosurgery. I must also determine how that will impact end users, such as physicians and patients.
My research in this field is very crucial because it really impacts millions of people, who may take AI answers as the ground truth when, in fact, they could be spreading misinformation about medicine. Making sure that such models do not spread information that is detrimental to neurosurgery and healthcare is so very important.
Another project that I’m really proud of uses AI to predict the probability of success at each step of a surgical procedure based on different brain signals that we collect through the course of the treatment and surgery. This will not only enhance the real-time alerting for surgeons during procedures but also provide important post-surgery insights and guide post-operative care with better precision.
What is on your horizon for other projects?
In terms of upcoming projects, there are several exciting avenues I’m exploring. As an expert in NLP, I’m particularly intrigued by the rapid advancements in generative AI and large language models, and I’m eager to incorporate them more extensively into my work.
I’m also keen to develop projects centered around the application of AI for social good. I am passionate about leveraging AI to make a meaningful impact in sectors that directly affect people’s lives. In addition to healthcare, I plan to explore the intersection of AI with agriculture and ethics, as AI has the potential to drive positive change in these areas.
I have the privilege of developing products using AI, a technology that can foster widespread positive impact. Each time I take on a new project, my overarching hope is that the solutions I craft will ultimately benefit individuals and communities, contributing to a more equitable future for everyone.
For more information about Sumedha Rai or to connect, please reach out to her on LinkedIn. Check out her session on NLP here.
