Chuong Ngo

I architected and deployed high-precision AI solutions for emergency response and dispatch environments, focusing on the intersection of technical performance and operational reliability. By engineering specialized ASR pipelines for slang-heavy dialects and high-noise audio, I delivered a 49% to +50% increase in transcription accuracy.

Beyond core transcription, I optimized real-time resource allocation to boost processing accuracy by 32% while maintaining strict computational constraints. My work includes overhauling speaker-attribution systems, increasing reliability by 42%, and establishing a responsible AI framework to eliminate hallucination risks. These interventions transformed raw, complex audio into a high-utility asset for mission-critical workflows.

I significantly enhanced the access and authorization systems for protected health data at the Centers for Medicare and Medicaid Services. By identifying and remedying critical vulnerabilities, I safeguarded personal health information from unauthorized access while ensuring it remained readily accessible for legitimate use. My work not only protects individuals from identity theft but also empowers authorized personnel with access to essential data to support the healthcare needs of those in their care.

Moreover, I transformed health data discovery with the QDAS Portal. The Portal's modern interface and refined user experience enable healthcare officials to quickly and effortlessly locate the critical information they need to perform their jobs efficiently.

I enhanced several of Alexa's NLP, ASR, and AI models and, in the process, uncovered specific performance deficiencies. By working closely with stakeholders both within and outside my team, I successfully addressed these issues, resulting in a significant 16% improvement in the models' sentence error rate.

I leveraged LLMs to build Vivid Health's healthcare transcription and insight-generation system. The system transcribed conversations between healthcare workers and their patients, saving them hours per week. It also generated summaries of the discussions that highlighted the essential topics and points. The summaries provided healthcare workers with invaluable insights that dramatically improved patient care quality.

I also created an AI-powered voice agent for healthcare debriefings. The solution enabled healthcare workers to focus on providing excellent patient care by simplifying administrative tasks.

I spearheaded the implementation of an advanced trust-based security system within a big data blockchain. This cutting-edge system incorporated the latest research on computational trust, enabling us to swiftly identify compromised network nodes and data integrity breaches. Additionally, it empowered us to adapt data and traffic routing, effectively countering network contagion.

I spearheaded the creation of the wildly successful Appellant Portal web app and back-end systems for the Center for Medicare and Medicaid Services. This groundbreaking project revolutionized the Center's health plan appeals process by introducing intuitive interfaces, top-tier security, and robust data processing. My efforts led to an impressive 94% reduction in the appeals process, from 100 days to seven days.

I was vital to bringing Woobo to life. Woobo was a revolutionary AI-powered educational toy. My contributions spanned the development of content recommendation, speech recognition, NLP, authentication, and other back-end systems. Additionally, I led the meticulous development of the Woobo toy, ensuring its top-notch quality through rigorous testing, automated code quality enhancements, and optimized database and data retrieval systems.

When researchers across the country needed secure access to sensitive CMS healthcare data, the existing pipeline was bogged down by legacy bottlenecks and friction. I stepped in to overhaul the entire web application and back-end platform, transforming a slow, cumbersome system into a modern, high-performance engine.

By migrating the core architecture to Next.js, React, Vite, and Mantine UI, I eliminated 10+ critical security vulnerabilities and met strict Section 508 and HCD compliance standards. To untangle data governance, I built a self-service RBAC gateway, upgraded the Python and PostgreSQL natural language analysis pipeline, and designed a unified tenant management framework. That shift completely changed the game: data-sharing lifecycles plummeted by over 90%, dropping approval and access times from weeks to mere minutes.

Beyond feature delivery, I locked down system reliability. When emergency outages struck, I orchestrated rapid service restoration, abolished obsolete auth caps, and codified operational workflows. Those fixes didn't just stabilize the platform—they drove an 80% reduction in MTTR, condensing days of troubleshooting into hours, and accelerated developer onboarding by 300%.