Developed an exercise authoring system enabling clinicians to create personalized rehabilitation tasks remotely in augmented reality using Unity, facilitating immersive clinician-patient engagement on Meta Quest 3.

Developed a web application adhering to Electronic Medical Record standards using React for the frontend and Next.js and Node.js for backend server-side rendering and routing. Utilized Firebase for secure data storage.

Conducted a comprehensive literature review on rehabilitation interventions and asymmetric communication to inform the innovative system design spaces for healthcare technologies.

Applied full stack skills to deploy SaaS tools powered by Large Language Models (LLMs) such as GPT-4.

Developed iOS, Android, and web applications using React.js, Node.js, NestJS, or Flask to support young adults in story writing and combating procrastination for mental health. Leveraged MongoDB for secure user data storage.

Designed and conducted user studies. Co-authored 3 peer-reviewed papers published in conferences and workshops, including CHI Conference on Human Factors in Computing Systems, contributing to user experience research.

Conducted and analyzed over 100 empirical experiments using Python and SLURM optimizing multilingual translation models including mBART, mT5, M2M, XLM-R, and NLLB.

Achieved a 20% BLEU score natural language processing improvement for low-resource languages through model building and domain adaptation techniques such as Intermediate Task Transfer Learning.

Demonstrated product management expertise by overseeing the full lifecycle of multiple research projects, showcasing flexibility in adapting to evolving requirements and fostering collaboration across cross-functional teams.

Developed backend features for an enterprise-class distributed database system based on PostgresSQL, implementing core functionalities such as buffer pool management, crash recovery mechanisms, concurrency control, fault-tolerant system using C++ and C.

Implemented LRU cache, consistent hashing, static B-Trees, and an LSM-tree with Bloom filters to optimize efficiency, achieving a 150% increase in datacenter tpmC throughput, as validated through TPC-C benchmarking. Optimized memory system by minimizing disk I/O, reducing false sharing, and avoiding fragmentation.

Led a quality assurance group, designing and executing over 50 tests, including G-tests, mock tests, concurrency tests, and integration tests, to systematically simulate distributed node scenarios and proactively mitigate node- specific anomalies in a multi-node, multi-threaded architecture.

Utilized Kernel-based Virtual Machine (KVM) to simulate scalable, isolated environments for testing and optimization, leveraging kernel-level debugging processes and tools to refine performance.

Delivered a technical presentation on system architecture to an audience of 40+ engineers, effectively communicating complex DevOps principles and implementation strategies.