AI Engineer Consultant at SmartHub.ai

Built agentic AI system for edge security using LangGraph and LLMs to monitor operational health across 10,000+ IoT endpoints, automating anomaly detection/root cause analysis and reducing incident response time by 45%. Engineered scalable data pipelines using REST APIs to ingest device telemetry and parse enterprise InfoSec policies into structured representations, enabling context-aware agent decisions on a ground-truth knowledge base. Developed LLM-driven automation for ticket generation and remediation workflow, reducing manual triage by 45%.

AI/ML Researcher at Bay Area Rapid Transit (BART)

Analyzed 100k+ LOC legacy codebase using NLP and LLM techniques, generating hierarchical summaries, metadata, and documentation artifacts to support large-scale analysis of complex enterprise software. Conducted evaluation of LLM-generated documentation, analyzing failure modes (e.g., hallucinations, omission, inconsistency) and iterating on prompts and preprocessing to improve reliability under on-prem constraints. Generated architecture and dependency visualizations using LikeC4/Graphviz to support system-level analysis.

Software Engineer at Samsung Electronics

Developed 15 2D/3D Unity game prototypes in C# to stress-test Tizen OS, benchmarking 10+ performance metrics (RAM, FPS, TTFF); isolated bottlenecks causing 30% latency spikes and defined strict asset budgets. Engineered automated deployment and testing pipelines using SDB CLI scripting, reducing manual effort by 50% and increasing daily data collection throughput from 15 to 80+ iterative builds for performance monitoring. Built profiling systems with configurable load injection to monitor metric impact on frame-time performance.