Applied AI Summit Healthcare
Free online conference | April 14-15, 2026Autonomous Cognitive Navigation: NLB Generative Frameworks for Scalable E2E Testing
Modern software validation is in crisis. Traditional automation frameworks built on hardcoded locators and deterministic scripts consume up to 70% of QA engineering time in pure maintenance, stifling innovation and slowing deployment velocity across an increasingly complex digital landscape.
This talk introduces Autonomous Cognitive Navigation (ACN), a Natural Language Based (NLB) generative framework that replaces brittle, script-driven testing with intent-based reasoning. At its core is a State-Aware Reasoning Loop powered by a Large Language Model that interprets real-time UI state, maintains a full chronological action history, and resolves high-entropy flows spanning 20+ steps, all from a single natural language instruction such as “Verify a user can successfully submit an insurance claim with a photo attachment.”
Three novel technical contributions drive the framework’s performance: a Persistent Temporal Action Memory that eliminates infinite-loop failures common in prior embedding-based AI systems, a Semantic Canonicalization function that reduces raw UI view hierarchy token volume by 85% for faster and more accurate reasoning, and a multimodal end-state verification engine that evaluates success visually, making it resilient to UI redesigns that break traditional identifier-based checks.
The results speak for themselves. ACN achieves a 95% reliability rate across thousands of localized application variants and 70+ languages, compared to near-100% failure rates of legacy scripts after a UI update. Onboarding new test flows drops from 120 hours of manual scripting to under 4 hours of natural language definition. At enterprise scale, this translates to an estimated 27 developer-years of saved maintenance effort and a 15-20% increase in feature release velocity.
This session is relevant to engineering leaders, QA practitioners, and product teams navigating the scalability demands of modern CI/CD pipelines across global.
About the speaker
Sowjanya Puligadda
Senior Engineering Manager at Uber
Sowjanya Puligadda is a Senior Engineering Manager at Uber with over 12 years of experience leading mission-critical engineering organizations across Uber and Amazon Web Services (AWS). She specializes in building and scaling global safety infrastructure, cloud ecosystems, and AI-driven quality platforms that support millions of users worldwide. Her expertise spans engineering leadership, distributed systems, deployment safety, and the application of Generative AI to software resilience. In her current role at Uber, Sowjanya serves as the principal architect of the company’s End-to-End Testing Northstar, a multi-year engineering vision governing software validation across Mobility, Delivery, and Freight. She has led high-stakes regulatory compliance programs, including the Context Propagation initiative that secured Uber’s legal right to operate in London, achieving zero tech-induced regulatory breaches in 2025 and 97% compliance across hundreds of PSD2 and TfL services. She also spearheaded DragonCrawl, an AI-driven mobile testing framework that detected over 230 critical bugs, prevented an estimated $2.5M to $10M in revenue losses, and saved more than 30 developer-years in manual effort. Additionally, she led the validation of UberVault, Uber’s ransomware recovery system, verifying the integrity of over 4 petabytes of data across 7,500+ resources. Prior to Uber, Sowjanya held an engineering leadership role at Amazon Web Services, where she directed the end-to-end global launch of AWS Lightsail, Amazon’s virtual private server platform. She built the foundational infrastructure, including Object Storage and Managed Databases, that enabled AWS to enter the individual developer and small business market. Her work delivered measurable customer impact, including a 40% cost reduction for migrating businesses and a dramatic reduction in setup time from 8 hours to just 15 minutes for SaaS providers. Earlier in her career, Sowjanya served as a Lead Software Engineer at Intuit, where she led full-stack development and Agile teams to build a government tax artifact acquisition system from the ground up, earning both the “Learn Fast” Spotlight Award and the prestigious Scott Cook Innovation Award for her revolutionary testing platform for tax consultants. She began her career as a Software Engineer at Infosys Technologies in India, focusing on identity management and backend development. Sowjanya holds a Master of Science in Computer Science from The University of Texas at Dallas and a Bachelor of Technology in Computer Science from Jawaharlal Nehru Technology University in India. She has published research on scaling mobile chaos testing with AI-driven test execution at ICSE 2026 and is a featured author on the Uber Engineering Blog. She leads multi-national teams across the United States, India, and the Netherlands, with a strong focus on talent development and operational excellence.