San Jose, California
ML engineer building agentic AI and personalized, on-device intelligence. Architected HPE's Aruba Networking Copilot — a hierarchical multi-agent system serving 6M+ devices, featured in the HPE Discover CEO keynote. Now at Google, and independently researching hyper-efficient small language models.
Featured Work
with Agentic Mesh Architecture
Led development of an industry-first autonomous network operations solution, featured in CEO Antonio Neri's keynote at HPE Discover 2025. A production-scale hierarchical multi-agent system that enables networks to think for themselves—autonomously identifying problems, offering solutions, and getting smarter over time.
Custom Neural Architecture for Network Intelligence
Engineered a custom Multi-Task BERT architecture that jointly optimizes ranking, intent/entity tagging, and snippet selection in a single forward pass. Optimized for CPU-only inference with ultra-low latency for real-time enterprise search.
Experience
Building personalized user-context and memory systems for Android & Chrome — context/memory architecture and efficient on-device intelligence.
Architected production-scale multi-agent systems using LangGraph. Key contributor to the HPE Aruba Networking Copilot featured in the HPE Discover 2025 CEO keynote. Built distributed AI infrastructure serving 6M+ devices and 3B+ clients globally with 50% MTTR reduction and 88% CSAT.
Engineered custom Multi-Task BERT architecture for enterprise search (Patent Pending). Built secure RAG pipelines with dimensionality compression. Enhanced AI-driven search achieving 98.5% accuracy with BERT-based algorithms and custom LLMs.
Led big data initiatives processing 800M+ records with Spark/HDFS, cutting latency by 15%. Implemented embedding-based semantic anomaly detection for network traffic, transforming 2B+ application flow records into vector spaces for real-time threat detection.
Developed StyleGAN-based generative models for 3D brain MRI achieving >95% IOU. Co-authored 6 peer-reviewed publications on deep learning for medical imaging. Built Auto-QC framework for semantic segmentation quality control.
Shipped AI at edge for digital pathology. Developed GAN-based image enhancement for whole slide imaging microscopy achieving >90% accuracy (Patent Pending). Deployed ML algorithms to AWS Cloud with 35% execution speed improvement.
Innovated advanced image processing algorithms for digital microscopy. Achieved 40% reduction in z-axis focus time through embedded programming for precision measurement.
Career Impact
Expertise
Research
Hackathons
3 hackathons. 3 wins. Rapid prototyping meets real-world impact.
Health analytics platform integrating wearable data with HL7-FHIR standards. Turns activity and mood data into actionable insights for better mental health outcomes.
Computer vision IoT device for food sustainability tracking. Monitors fridge contents to reduce food waste and promote sustainable consumption habits.
Real-time American Sign Language translation system using Transformer architectures. Enables accessible communication during video calls.
When I'm not building AI systems, you'll find me chasing light in national parks. The patience required to wait for the perfect shot translates directly to debugging complex systems—both reward those who observe carefully and act decisively.
