Hello There!
I am a computational researcher with proven experience across machine learning, clinical language models, agentic systems, and on-device optimization. At the University of Colorado Anschutz (The Mitchell Lab), my work centers on human-centered clinical AI, designing privacy-preserving architectures that bring language models directly to edge devices under HIPAA constraints, and engineering autonomous multi-agent frameworks that automate requirement-elicitation and patient simulations.
Prior to my postdoctoral fellowship, I served as a Staff Researcher and AI Camera Researcher at Lenovo Research, delivering mission-critical Gen-AI platform features (gaze and blink correction), ultra-lightweight super-resolution modules (8ms latency), and novel GAN training pipelines with automated CI/CD benchmarking.
I completed my PhD in Electrical and Computer Engineering at Binghamton University under Dr. Xiaohua Li, investigating adversarial defense mechanisms and query optimization in deep learning. I thrive at the intersection of rigorous scientific research and production deployment.
Recent News
2025 - 2026Key Research Projects
Flagship architectures across clinical AI, agentic systems, and computer vision
PLACID: Privacy-Preserving Clinical Acronym Disambiguation
Architected a two-stage cascade routing general-purpose instruction models for acronym detection to domain-specific biomedical models for context expansion. Benchmarked on local small models (2B–10B) achieving ~0.988 detection accuracy and raising expansion accuracy to ~0.81 with zero external data egress.
LadderTeam: Dual-Agent Elicitation Framework
Architected a dual-agent (Interviewer & Judge) framework automating laddering interviews (ACV, 5-Whys, JTBD). Built a reproducible 5-step per-turn loop with drift detection and non-interrupting evaluation judging. Achieved 99.1% chain convergence and zero drift across 216 controlled simulation runs.
Enterprise Gen-AI Camera & Model Optimization
Engineered generator architecture for enterprise Gen-AI (blink and gaze correction) with lightweight 8ms Super Resolution module (+71MB GPU). Designed scalable novel GAN CI/CD infrastructure improving training efficiency by 40% and cutting manual QA by 85%. Led A/B testing dashboards that cut testing cycle from 6 to 2 hours.
Mitigating Adversarial Attacks in Deep Learning Systems
Investigated adversarial vulnerability and developed defense algorithms against state-of-the-art black-box adversarial attacks, achieving 100% attack mitigation rate for critical applications. Implemented evaluation pipelines using Cleverhans and ART in PyTorch/TensorFlow, and researched pretrained VAE and DCGAN architectures for query optimization.
Experience & Background
Chronological summary of research appointments, industry leadership, and academic instruction
Postdoctoral Fellow: Clinical AI Framework Development
CurrentDeveloping privacy-preserving clinical language model architectures and autonomous multi-agent requirement elicitation frameworks for healthcare systems under strict HIPAA and data-privacy constraints.
Staff Researcher & AI Camera Researcher
Nov 2022 – Oct 2025Led enterprise Gen-AI camera feature development (blink and gaze correction), lightweight super-resolution modules (8ms latency), novel GAN training CI/CD pipelines, and PEFT/LoRA mobile deployment optimization across Qualcomm and MediaTek architectures. Recipient of Lenovo SVP Individual Excellence Award.
Graduate Research Fellow: AI Robustness & Security
Aug 2019 – Oct 2022Researched adversarial vulnerabilities in deep neural networks and developed defense algorithms achieving 100% attack mitigation against black-box adversarial attacks. Implemented evaluation suites with Cleverhans and ART toolboxes in PyTorch and TensorFlow.
Teaching Assistant: Engineering Design Division (EDD)
2019 – 2022Taught undergraduate courses in signal processing and mentored senior capstone engineering design projects over 3 years. Developed instructional curricula covering coding, CAD, and circuit design; awarded Binghamton Teaching Excellence Award.
Hardware Engineering Research Intern (Co-op)
Jan 2018 – Jun 2018Developed high-speed wafer feeder systems using Xilinx platforms with precision spindle switching mechanisms. Programmed embedded code for FIFO and power control systems, and implemented digital filters for high-speed noise cancellation.
Education
Publications & Presentations
All peer-reviewed journal papers, conference proceedings, preprints, and symposium posters