I lead architecture and development while remaining a primary hands-on contributor across enterprise GenAI, knowledge graph, MCP, administration, and billing platforms. The common challenge is turning information and actions scattered across systems into governed, production-ready tools that people can use and trust.
- Owned the end-to-end automation of work connected to the federal Rural Digital Opportunity Fund (RDOF), a program supporting broadband expansion. I connected PostgreSQL and Teradata data with forms, ticketing, approvals, and reporting so the process no longer depended on a long chain of manual handoffs. Turnaround fell from roughly nine weeks to two, and the system handled about 400,000 serviceability passes in one quarter.
- Contributed to a questionnaire and offer-generation platform that turns a complicated set of inputs—customer type, location, available products, valid combinations, and local business rules—into eligible offers. The goal was to reduce a process that could take far longer into one that operates on the scale of hours.
- Led architecture and development as a primary contributor for a multi-tenant GenAI platform powering 7+ production AI assistants. I built React and TypeScript experiences plus Python services with RAG, LangChain and LangGraph orchestration, streaming conversations, memory, and dynamic LLM tool calling.
- Engineered hybrid retrieval across FAISS, BM25, PostgreSQL and pgvector using Reciprocal Rank Fusion, S3-synchronized indexes, metadata matching, and document-context reconstruction. Structured-data providers included governed Snowflake text-to-SQL workflows with validation, retries, and controlled execution.
- Led and personally implemented an ontology-driven Neo4j knowledge graph covering people, billing, projects, Jira work, source code, and other core enterprise entities. I designed config-driven AWS Glue and Spark pipelines with normalization, validation, idempotent loaders, and cross-system identity resolution.
- Initiated, architected, and independently built a React, TypeScript, and FastAPI administration platform for Kubernetes, EKS, and multi-account AWS Glue operations, with OAuth and SSO, JWT authentication, fine-grained RBAC, automated token lifecycle management, and AI-assisted root-cause analysis.
- Architected and led a governed MCP gateway integrating Jira, Neo4j, Teradata, SQL, and enterprise billing systems, with per-consumer authorization, rate limiting, Redis caching, audit logging, KMS encryption, Secrets Manager, and Kubernetes deployment.
- Led Java 17 and Spring Boot 3 services plus enterprise billing and data workflows spanning 4.1M+ records across 141 code sets, integrating Snowflake, SQL Server, Teradata, DocumentDB, and S3.
ReactTypeScriptPythonFastAPIJava 17Spring Boot 3LangChainLangGraphRAGAI AgentsFAISSBM25RRFPostgreSQLpgvectorSnowflakeSQL ServerTeradataDocumentDBNeo4jCypherOntology ModelingAWS GlueApache SparkAmazon S3AWS KMSAWS Secrets ManagerMCPOAuth 2.0SSOJWTRBACRedisJiraDockerKubernetesHelmAmazon EKSGitLab CI/CDDatadog