Kevin Lee

Led a team of 6 engineers to build a C#/.NET outage risk forecasting platform processing 1.2M+ daily utility records, integrating SageMaker ML models to improve prediction accuracy by 27%. Built an AWS Bedrock/LangChain/Pinecone RAG system enabling field engineers to query 50K+ documents and reduce issue resolution time by 35%.

Delivered a production Azure RAG pipeline using Azure OpenAI and LangChain across 500K+ compliance documents, improving answer accuracy by 30% and reducing analyst review time by 3 hours/day. Built multi-turn conversational AI agents serving 10K+ daily active users with sub-2s p95 latency and a 99.5% SLA for ML inference services on AKS.

Built Databricks ETL and ML feature pipelines processing 5TB+ of enterprise data daily and improved GPT-based text labeling accuracy by 30% across 3 enterprise clients. Migrated 12TB from on-prem SQL Server to Azure Data Lake/Databricks Delta Lake using a medallion architecture, and delivered streaming pipelines processing 100K+ events/min for dashboards.

Built an AI document analysis platform from scratch processing 1M+ financial documents, reducing manual extraction time by 35% for enterprise clients. Trained and deployed Transformers for named entity recognition and clause extraction (92% F1) and implemented real-time semantic search with Elasticsearch/vector embeddings and Redis caching, reducing p95 query latency from 1.2s to 240ms.

Built and optimized CMS APIs with Django REST Framework and Node.js for 500K+ daily requests, reducing p95 response time by ~40% via query tuning and connection pooling. Modernized legacy PHP/Laravel services to Node.js/Express/MongoDB and reduced API latency and dashboard query time by 60% while cutting MTTR from 4 hours to 45 minutes through observability improvements and SOC 2 alignment.

Created a Vue.js customer support automation UI for 200+ support agents, improving CSAT by 20% with streamlined ticket workflows and real-time status updates. Built a React/TypeScript monitoring dashboard using GraphQL subscriptions and integrated Kafka streaming with sub-500ms end-to-end latency, while increasing test coverage to 85%+ and reducing production bugs by 40%.

Built Python/Boto3 automation for AWS EC2 and S3 resource management, reducing manual provisioning effort by 30% and improving incident response workflows with CloudWatch monitoring dashboards and alerts. Developed interactive onboarding tutorials in HTML5/CSS3/JavaScript, increasing new user completion rate by 20%.

Earned a Master of Science degree at San Jose State University from 2013 to 2015.

Earned a Bachelor of Science degree at San Jose State University from 2009 to 2013.