David Adams

Helped deliver the MyCity platform, supporting 125,000+ users accessing municipal services with an emphasis on reliability, secure workflows, and production readiness. Built and supported an AI conversational assistant using Python, FastAPI, LangChain, and OpenAI, reducing Tier-1 support tickets by 35%. Designed AI-driven workflow orchestration for request classification, retrieval, and task execution with policy controls and traceable processing. Developed an AI document-verification workflow that reduced permit processing time from five business days to under four hours. Implemented semantic search across two million civic records and optimized backend processing for stronger query performance and API throughput. Built reusable React/Next.js components and integrated GraphQL across multiple services to improve UI consistency and simplify data access. Supported Azure application operations, identity/secrets practices, monitoring, incident triage, and controlled releases across environ

Delivered full-stack features and Python data-processing workflows for SmartTrials, improving throughput across life-sciences research operations. Integrated NLP-based extraction and LLM-enabled workflow capabilities into biomedical research processes, reducing dataset onboarding time by 60%. Built and maintained secure service integrations and data pipelines supporting regulated workflows and reliable access to research information. Improved cloud deployment practices using Azure/AWS, containers, Kubernetes, and CI/CD controls for applications serving 2,000+ research users. Optimized processing pipelines and model-serving efficiency with dynamic batching and model quantization, reducing infrastructure costs by 30%. Worked within change-controlled delivery practices including testing, peer review, release coordination, monitoring, and production support.

Built scalable REST APIs and backend data pipelines for clinical platforms processing multi-terabyte datasets across 40+ client environments. Automated testing and data-validation workflows, reducing release verification time from 14 days to eight hours. Developed data-quality and anomaly-detection pipelines that reduced manual reconciliation and improved trust in clinical data. Implemented asynchronous processing with RabbitMQ to reduce synchronous bottlenecks and improve reliability. Applied encryption, role-based access control, audit logging, and controlled access practices to support HIPAA- and FDA-regulated data handling. Investigated production issues, collaborated with cross-functional teams, and delivered maintainable fixes through test and release workflows.