Bright Wang

Led development and production deployment of LLM and ML pipelines emphasizing reproducibility, observability, and developer ergonomics; shipped RAG pipelines that improved workflow efficiency by 32% and implemented safety guardrails and evaluation frameworks for regulated environments.

Advised and prototyped production-ready GenAI systems focusing on performance, evaluation, and reliability; built RAG evaluations, agent orchestration prototypes, and benchmarking harnesses to measure latency, throughput, and quality trade-offs.

Built scalable ML platforms and cloud-native AI services, designing distributed training pipelines for 100M+ records and prototyping optimized inference stacks with Triton to improve latency and deployment portability.

Applied ML research in NLP and computer vision, developing NER and OCR pipelines and supporting GPU-based PyTorch training and reproducible experimentation during M.S. studies.

Developed and productionized real-time ML systems for fraud detection, improving model precision by 22% and delivering low-latency inference services under 100ms with automated monitoring for drift.

Prototyped and evaluated NLP classifiers for medical document classification, building scikit-learn models, conducting exploratory analysis, and supporting annotation and reranking experiments.

Completed a Master of Science in Computer Science while conducting research and applied work in NLP and computer vision as a graduate research assistant.

Completed a Bachelor of Science in Statistics with coursework and practical projects supporting data analysis and foundational machine learning concepts.