In an interview with TimesTech, Meetu Malhotra, Data Analytics Principal at S&P Global, shares her insights on uniting academic research with practical AI applications. She discusses trends in generative AI, the growing focus on domain-specific solutions, and the role of mentorship in tech. Meetu also talks about her books on cyberbullying and machine learning, emphasizing responsible AI, inclusivity, and the translation of innovation into tangible business impact.
Read the full interview here:
TimesTech: With over 17 years of industry experience and a strong academic foundation, how do you bridge the gap between theoretical research in data science and its practical application in your role as a researcher and Data Analytics Principal?
Meetu: My 17+ years of industry experience and strong foundation in data science enable me to connect academic insights with real-world solutions, ensuring that the projects I work on are technically sound, scalable, and aligned with business needs. As a researcher, I stay deeply engaged with the latest innovations by regularly reading academic papers and books. This continuous learning keeps me current with emerging technologies and methodologies, which I then apply both in my research and on the job. It enables me to translate abstract concepts into real-world solutions by bringing cutting-edge developments directly into production environments and solving business-critical problems with informed, up-to-date knowledge. For example, staying current with generative AI research allowed me to apply it to a use case that meaningfully enhanced data quality. Additionally, collaborating with cross-functional stakeholders ensures that the work stays relevant, innovative, and rooted in solving actual organizational challenges. I believe the real power of data science lies not just in applying cutting-edge methods but in translating them into solutions that make measurable and meaningful business impact and that’s the bridge I strive to build every day.
TimesTech: You’ve reviewed over 150+ book chapters and papers for reputed journals and conferences, including ICML and IEEE Access. What key trends or research gaps do you commonly observe in the current landscape of AI and data science?
Meetu: Since the publication of the paper “Attention is All You Need” (2017), which introduced the Transformer architecture, we’ve witnessed a tsunami of innovation in transformer-based models. Initially applied to text processing tasks, transformers have now expanded across nearly every domain and every data type. For example, models like TimeGPT have emerged, leveraging transformers for time series forecasting, WaveNet for audio processing etc. With this, one of the most prominent trends I’ve observed is the explosive growth in generative AI and large language models (LLMs). Researchers are actively working to enhance LLMs by refining architecture, increasing training efficiency, reducing computational costs, and addressing ethical, safety and environmental concerns. A key direction now is the application of LLMs to domain-specific problems. For instance, detecting emotion in regional languages, analyzing sentiment in social media discourse, or using hybrid approaches for abstractive text summarization. Additionally, the field is beginning to pivot toward Small Language Models (SLMs) that are more efficient, multimodal such as Visual Language models and that show a new wave of innovation focused on scalability and sustainability.
However, one key gap I notice is the lack of focus on domain adaptation, meaning many papers emphasize model performance but fall short on acknowledging that these models cannot generalize across real-world scenarios. Another recurring issue is the challenge with reproducibility as many submissions lack accessible code, and clear experimental steps. On the positive side, interdisciplinary research, which means combining AI with social analytics, healthcare, agriculture, education, law, finance, cybersecurity and almost every domain is gaining momentum, which is encouraging. As a reviewer, I consistently advocate for research that is not only novel but also actionable, transparent, and socially responsible. I believe these are the pillars on which the future of meaningful AI research will be built.
TimesTech: Your recent book with IGI Global on combating cyberbullying using GenAI touches on a socially crucial topic. What inspired you to explore this area, and what role do you see GenAI playing in digital ethics and online safety moving forward?
Meetu: The inspiration behind my recent book “Combating Cyberbullying with GenAI” published by IGI Global stemmed from both professional curiosity and personal concern for the growing mental health crisis among youth and digital users. Cyberbullying is not just a technological issue rather it is a deeply human one with long-lasting emotional and psychological effects. According to a study at cdc website, teenagers who experienced bullying were almost twice as likely to report symptoms of anxiety (29.8%) or depression (28.5%) compared to those who were not bullied. As a researcher in AI, I felt a strong responsibility to contribute towards using emerging technologies like generative AI for socially beneficial purposes. GenAI offers immense potential in flagging harmful content and generating safe and inclusive communication in real-time. However, there are indeed some challenges associated with GenAI around its misuse involving ethical concerns, deepfakes, and bias because LLMs are trained on data which mirrors biases in our society. My book addresses these dualities—how to leverage GenAI as a force for good while embedding ethical safeguards and regulatory considerations. I foresee GenAI playing a pivotal role in shaping digital space for good. Through this work, I hope to spark deeper conversations and solutions at the intersection of AI, digital ethics, and societal well-being, encouraging more researchers and technologists to take up cause-driven innovation.
TimesTech: You are actively engaged in public speaking at major platforms, including the 2025 Global Women in Tech Conference. How important is representation and mentorship in tech, especially for women and underrepresented communities in AI/ML?
Meetu: Representation and mentorship in tech, especially in emerging fields like AI/ML are essential to creating inclusive innovation. I have had the privilege of speaking at influential platforms such as the 2025 Global Women in Tech Conference and Data Analytics Summit. Such platforms provide opportunities to showcase the unique perspectives and inspires others in similar situations to pursue careers in tech. Women and minorities often face unique barriers in accessing opportunities, gaining recognition, or advancing in STEM careers. I believe mentorship is one of the most powerful tools to bridge this gap. I actively support emerging talent through initiatives like the STEMUp mentorship program and Girls Who Code career mentoring session which particularly focus on individuals who may not have traditional pathways into tech. By providing guidance and resources, I aim to empower these individuals to navigate their careers in technology. This commitment to mentorship encourages innovation by bringing fresh perspectives into the industry. My own journey involves balancing industry roles, academia, and research and has been shaped by the guidance of my professors and mentors. I simply view it as my responsibility to pay that forward.
TimesTech: As someone with hands-on experience in reviewing technical books and articles, how do you ensure quality and relevance in a rapidly evolving field like AI, where breakthroughs happen almost weekly?
Meetu: The dynamics of AI are changing rapidly with new models and frameworks emerging almost weekly. Therefore, it’s essential to stay updated. To enhance my expertise, I continuously equip myself with the latest use cases and challenges in AI, staying informed through reliable blogs and books. This commitment to ongoing learning enables me to ensure that the material I review is both reliable and effective in deepening the reader’s understanding of complex subjects. As a certified reviewer with Elsevier and ACM and having reviewed technical books for publishers like O’Reilly and Manning and research papers for many reputed conferences and journals, I’ve learned to assess not just technical accuracy but also educational value. Cross-referencing information with other reputable sources is essential for validating accuracy. I also prioritize the clarity of the writing, ensuring that complex concepts are accessible to the intended audience to help with clear explanations that enhance understanding. Furthermore, I pay particular attention to emerging trends, such as responsible AI, to ensure that the content is not only relevant but also aligned with current ethical considerations in the field. I also emphasize reproducibility in the submitted papers, meaning if the authors have provided the datasets, code snippets, or GitHub links. I believe that reviewers serve as guardians of both knowledge and trust, and it’s a responsibility that I take seriously to ensure that the readers and learners are equipped with the most credible and relevant material.
TimesTech: Your upcoming book on machine learning sounds promising. Could you give us a preview of what readers can expect and how you plan to balance depth, accessibility, and real-world application in your writing?
Meetu: My upcoming book on machine learning is designed to serve as both a practical guide and a conceptual deep dive into the core principles that underpin modern ML applications. Readers can expect a structured journey starting from the basics—data preprocessing, feature engineering, and exploratory data analysis—towards more advanced topics such as ensemble models, time series forecasting, and recommendation models. What sets this book apart is its ability to break down the mathematics behind each algorithm using small datasets while also explaining the concepts in clear, accessible language. It blends theoretical rigor with practical application, featuring real-world case studies drawn from diverse research projects. Each chapter includes well-structured Python code examples to help readers implement concepts step by step. Additionally, the book features concise “Notes” sections within chapters, set to clarify common misconceptions, and prepare readers for job interviews by highlighting practical tips. This thoughtful structure ensures that learners not only understand the ‘how’ but also the ‘why’ behind machine learning techniques. The goal is to empower readers to not just use ML libraries, but to think like data scientists and make informed decisions. I want this book to resonate with students, early-career professionals, and even experienced analysts looking to deepen their understanding of the ML concepts. It will be published soon.
