Apoorva Kasoju is using her knowledge and experience to change the way we use artificial intelligence in real-time applications and improve the way these technologies can help and impact people in their lives.
AI and machine learning models have advanced from research labs to being commonplace in many products we interact with daily. Recommendation engines, spam filters, anomaly detection systems, real-time diagnostics and many more real-time AI applications are an integral part of our modern world. As these models grow in complexity, so too do the calls for transparency, ethical uses and a demonstrable human impact.
Apoorva Kasoju, Senior Software Development Engineer at Amazon, is one of the many people helping to change what it means to build intelligent systems that are both technically excellent and socially responsible.
After working in the technology industry for over nine years, Apoorva Kasoju has become an expert in building cloud-native software and AI systems that scale to meet the needs of millions of users. Her experience working on mission-critical cloud infrastructure for Amazon and pioneering cutting-edge technical projects has taught her the value of building AI systems with accountability. At Amazon, Kasoju’s leadership in the FBA Dashboard and Customer Returns Card projects led to increased seller engagement and return reductions, saving the company hundreds of millions of dollars in cost. However, her contribution has not only been in writing algorithms that improve metrics but also in building systems that meet regulatory compliance and perform well across diverse languages, markets, and regions.
Kasoju has also made a name for herself in transforming legacy systems into real-time data processing pipelines and using serverless workflows to redesign business processes. In one of her recent cloud-native software engineering projects, she helped re-architect a system for law enforcement agencies to transition from a manual process for management of crime cases to a modern and event-driven architecture. The complete re-architecture of the system from a monolithic application, hosted on physical servers to a serverless event-driven architecture, was achieved using AWS Step Functions and Lambda.
The newly built software helped in faster and more secure processing of a few thousand cases every month while scaling seamlessly across continents to benefit law enforcement globally. For Kasoju, projects like these are an embodiment of the same philosophy. Impact comes from architecture, but it is always grounded in a practical understanding of user experience.
Her technical expertise and engineering prowess are complemented by her active research on developing more explainable and accountable AI systems. In one of her recent publications, Kasoju and co-authors have demonstrated how explainability techniques, such as SHAP, can assist human clinicians in evaluating and validating ML models for automated, objective skin cancer diagnoses. In her view, innovation should not come at the cost of clarity or trust. “Especially in cases when the decision might impact a person’s health or livelihood, innovation is only meaningful if it can be trusted,” Kasoju said.
Her recent work on reducing latency and cost of transformer models at inference time has also highlighted the rising importance of striking the right balance. Using model optimization techniques like quantization, precomputed embeddings, and model distillation, Kasoju’s team at Amazon has demonstrated real-time prediction speeds that were considered impossible just a few years ago. Kasoju sees new edge deployments, dynamic request routing, and hybrid inference as the next big opportunities in AI systems. By using these approaches, it will be possible to leverage powerful language models while reducing cost and latency for users.
Away from her contributions in the tech industry, Apoorva Kasoju has also built a name for herself as a mentor and community influencer. Her work as a judge at Amazon’s innovation challenges and visiting speaker at several universities has seen her train and mentor many students and early-career engineers in translating research and academic ideas into working projects and industry-ready solutions.
Kasoju strongly believes that this mentorship and giving back to the community are not just a responsibility but an essential part of one’s growth. “I have learned more by teaching and guiding people than I thought I could,” she shared. Her mentorship efforts have also seen her get involved in grassroots efforts to support local women and girls to study technology by helping them access resources and opportunities.
In 2025, Kasoju was recognized and named the International Achievements Research Center’s (IARC) Technology Leader of the Year. The award recognizes Kasoju’s technical accomplishments in software development and AI engineering and acknowledges her significant contributions as a mentor to the next generation of computer scientists. Kasoju says that the award is as much about her personal milestones as it is about all the individuals and communities she has been privileged to support along the way.
AI technologies are becoming an increasingly integral part of our daily lives. However, as computing and models continue to improve, the impact of AI on our society and individuals is becoming clearer. It is also becoming apparent that leaders like Apoorva Kasoju are driving the AI industry in a way that is responsible, transparent, and, above all, kind. Success is no longer just a matter of how many requests per second a system can process or how accurate a model is.
It is also a measure of the care with which these systems are built and the positive impact these systems have on users. In an industry where the emphasis is on ever-greater speed, Kasoju’s life and work are a reminder that the true value of technology is the positive change it creates in the world.
