Intuit India recently hosted a highly anticipated woman at the Data Science Worldwide (WIDS) Bengaluru 2025 conference held on June 12th. The event, held at Intuit's Bengaluru office, served as an inspiring platform for experts to connect, learn and explore the future of AI and machine learning in an inclusive and dynamic environment.
As the global financial technology platform behind TurboTax, Credit Karma, Quickbooks and MailChimp, Intuit is committed to driving data science innovation. Started by Stanford, the WIDS Conference grew from a one-day event to a global movement, reaching one or more experts each year through conferences, datasons, podcasts, workshops and programs aimed at fostering the next generation of data scientists.
The event began with an opening statement from Intuit's WIDS brand ambassador Vignesh Subrahmaniam. Subramaniam highlighted the groundbreaking contributions of Florence Nightingale, the first woman to apply statistics and data science to improve healthcare, and her strong legacy of data-driven solutions.
Subramaniam's words set the tone of the meeting, reminding participants of the pioneering role women played in shaping the world of data science.
Insightful sessions and keynotes
The keynote session by Amruta Joshi, managing director of Google's AI Solutions, went deep into concrete innovation. We have clarified the role of embedded systems, delegation, and current limitations of AI. This session resonated deeply with participants who were encouraged to focus on not only AI potential but also practical and creative applications that could lead to real impact.
One outstanding session was led by Sridhar Dasaratha, Head of AI Research at EY. Dasaratha grounded the AI-generated answers to the actual data, minimizing hallucinations, and emphasized that generation only occurs when the model has authentic knowledge. This session sparked deep debate on the importance of data integrity and accuracy when using AI in high stakes fields, such as finance where accuracy is not negotiable.
Next, Jayashree Mohan of Microsoft Research took the stage to deal with the LLM serving and presented the challenges of system design. She discussed the latest advancements in Scalable training and efficient inference techniques such as Flashattention, VLLM, ORCA, and TokenWeave. Mohan also highlighted the important role of performance analysis tools such as the LLM Judge and LMSYS-chat-1M. Her message was clear. A robust infrastructure is key to achieving not only models but also reliable and scalable AI performance. Her sessions captivated the audience and highlighted the importance of scalability for real-world applications.
Microsoft Research's Anjaly Parayil emphasized that rather than being added as an afterthought, the fairness of AI systems should be integrated throughout the design process. She has introduced FairServe, a framework designed to ensure fair LLM access amidst competing demand. Using mechanisms such as overload-driven throttling, debt-based prioritization, and weighted service counters, FairServe ensures fair access in AI systems. The Parayil session sparked a lively debate on how to balance equity and efficiency in AI-powered systems.
Later that day, Intuit's Sid Kumar and Arkadeep Banerjee provided a simplified visual explanation of reinforcement learning from human feedback (RLHF). Banerjee further emphasized the context, relevance and cohort importance when processing data, and introduced the concept of data QNA.
He pointed out that while RLHF is a powerful method, there are scenarios where non-RLHF methods are more appropriate, especially when simple feedback is sufficient. Their session was a perfect blend of theory and practical applications, providing valuable insight into context-conscious AI systems.
The meeting concluded with a session by Niyati Chhaya, co-founder of Hyperbot. She highlighted the need to develop agent workflows in finance that takes into account unstructured data, trust establishment, and verification processes.
Chhaya's speech provided a roadmap for streamlining complex financial tasks, highlighting the need to design AI systems to handle the nuances and complexity of real data.
Day of inspiration and collaboration
The WIDS Bengaluru 2025 conference was a huge success, offering a rich mix of insights, discussion and networking opportunities.
The session features key topics ranging from Finance's LLM applications to AI fairness and system design, all contributing to the future of data science and AI. Participants felt inspired and empowered by pushing the boundaries of data science innovation. As Intuit continues to advocate for diversity and inclusion with AI, the WIDS Bengaluru 2025 conference is a testament to the growing role of women in shaping the future of technology. Stay tuned for the next edition as industry pioneers continue to share cutting-edge knowledge and thrive in the rapidly evolving landscape. Find out more at Intuit Careers.
