I build machine learning systems at the intersection of federated learning, health AI, and agentic AI. My research focuses on enabling privacy-preserving, distributed analysis of large-scale clinical and behavioral data — designing algorithms that work across heterogeneous sites without centralizing sensitive records.

Currently a Postdoctoral ML Researcher at Yeshiva University (XR AI Lab). PhD in Engineering & Applied Science (Computer Science), UMass Dartmouth, Oct 2025.

Research areas: Federated & distributed learning · Unsupervised/fuzzy modeling · Missing-data frameworks · Computer vision (BEV-map, V2V perception) · LLM fine-tuning & agentic AI · Health AI · Privacy-preserving ML

Under review · ACM Transactions on Computing for Healthcare

End-to-end federated clustering pipeline for incomplete multi-site longitudinal health data.

• Lossless FeFCM protocol preserving full fuzzy membership across sites
• Fuzzy co-association + NMF consensus aggregation: ARI 0.494 → 0.593 (free-K setting)
• GapDB composite automatic (K, m) selection without global data sharing
• Validated on 6 datasets including 2 national cohorts: 3M+ records, 88% structural missingness, 31:1 site imbalance

Manuscript in preparation

Federated framework where each site runs a QLoRA-finetuned LLM agent (Qwen-7B/72B) that autonomously selects its own clustering pipeline and cluster count K from a 49,920-configuration space.

• Trained on 1,915 expert trajectories across 246 datasets (fully held-out train/val/test splits)
• Two-tier orchestrator: method-specific federated protocol + trust-weighted EMA fallback
• Evaluated on 4 real MA RCT cohorts + MIMIC-eICU clinical held-out validation

Published · IEEE Transactions on Vehicular Technology (IF 6.1, 73 citations)

BEV-map based V2V collaborative perception for autonomous vehicles to resolve occlusion.

• Inference latency: 0.0456 s/frame (~21.9 Hz) — meets real-time 10 Hz requirement
• BEV map 1.1 Mb @ 0.011 s vs compressed LiDAR 4,000 Mb @ 40 s over 100 Mbps
• Tested on KITTI dataset

Published · IEEE Internet of Things Journal (IF 8.2, 14 citations)

Novel federated fuzzy clustering for decentralized incomplete longitudinal behavioral data.

• ~93% accuracy vs ~52% decentralized FCM baseline at 20% missingness (N=30k, 8 clients)
• Validated on 4 real clinical cohorts (N=957, 4 sites, NIH R01-funded)
• Converged in ~10–80 iterations; reproducible multi-trial evaluation harness

Under Review / In Preparation

• H. Ngo, H. Fang, H. Wang — FUSE: Federated Fuzzy Consensus Clustering Under Missing Data for Longitudinal Digital Health Trials. ACM Trans. Computing for Healthcare (under review).
• H. Ngo et al. — FedAutoCluster: Distilling Clustering Expert Judgment into Federated Per-Site LLM Agents. In preparation.

Journal Articles

• H. Ngo et al. — Federated fuzzy clustering for decentralized incomplete longitudinal behavioral data. IEEE IoT Journal (2023). IF 8.2 · 14 citations
• H. Ngo et al. — Cooperative perception with V2V communication for autonomous vehicles. IEEE TVT (2023). IF 6.1 · 73 citations
• S. V. Balkus, ..., H. Ngo et al. — A survey of collaborative ML using 5G vehicular communications. IEEE Comm. Surveys & Tutorials (2022). IF 35.6 · 128 citations
• B. Cornet, ..., H. Ngo et al. — An overview of WBANs for mobile health applications. IEEE Network (2022). IF 9.3 · 88 citations
• H. Ngo et al. — Beamforming and scalable image processing in V2V networks. J. Signal Processing Systems (2022).
• V. S. Gurugubelli, ..., H. Ngo et al. — A review of harmonization methods for studying dietary patterns. Smart Health (2022).

Conference Papers

• H. Ngo et al. — Deep learning-based adaptive beamforming for mmWave WBAN. IEEE GLOBECOM 2020. 🏆 Best Paper Award (IEEE ComSoc)
• H. Ngo et al. — Intelligent fuzzifier-based cluster validation. IEEE/ACM CHASE 2022
• J. Matos, H. Ngo et al. — XR-enabled digital twins in longitudinal trials. IEEE/ACM CHASE 2025

📚 Full list on Google Scholar · h-index: 8 · 382+ citations

Languages

ML / Deep Learning

Agentic AI / LLMs

Infrastructure

🏆 Best Paper Award — IEEE ComSoc Multimedia Communications TC, IEEE GLOBECOM 2020
🎓 CIS Graduate Research Award — UMass Dartmouth, 2023
✈️ NSF Student Travel Grant — IEEE/ACM CHASE, 2022
🔬 NSF Grants — ECCS 2010366, IIS 2140729 | NIH — R01DK129432
🏅 CMU × NVIDIA Federated Learning Hackathon — Team Lead, 2026 (privacy-preserving biobank AI)