• Developed full-stack features for enterprise educational services with React/Next.js frontend, authentication middleware, and RESTful APIs; created 20+ reusable components including infinite scroll with caching and role-based access control

• Collaborated with researcher to implement AI-powered vector search using embedding generation and Claude API integration, helping automate creation of intermediaries and through semantic similarity scoring

• Led technical demonstrations for stakeholder presentations to Tools SIG and Salience Learning teams

• Led team of 5 engineers across AI and software development, conducting weekly meetings and providing technical mentorship on cloud architecture; established technical vision with architectural diagrams presented to advisors and customers

• Architected and deployed Emporia, an AI agent using Claude API with function calling to intelligently route user requests between claims generation, PubMed natural language search, and context-aware question answering for competitive analysis

• Designed end-to-end document processing pipeline: Azure Document Intelligence OCR → LangChain semantic chunking → Pinecone vector embeddings, enabling LLM-powered marketing claims generation from uploaded research papers

• Migrated Docker-based Flask/PostgreSQL app to scalable microservices architecture managed via Terraform IaC, reducing monthly cloud costs by hundreds of dollars while improving maintainability and performance

• Built Flask backend with session-based authentication (Flask-Login), Azure CosmosDB NoSQL database, and Pydantic validation; developed React/TypeScript frontend to showcase AI capabilities

• Established engineering vision for the start-up and employed Agile methodology to roadmap software with business needs

• Managed team of 5 developers, establishing team-wide development standards and lead 1-1 check-ins and code reviews

• Developed responsive React web app to onboard new users with signup and login functionality and add users to a waitlist

• Set up AWS CDK infrastructure as code to improve maintainability, extensibility, and transparency for serverless backend

• Constructed circuit to induce controllable voltage spikes that allow organization of collected OpenBCI EEG time series data

• Pre-processed EEG Matlab data, filtered out noise frequencies with SciPy and split data into training and test sets, trained neural network classifier to predict whether the English sentence read by the observer was hard or easy to complete

• Wrote lesson about using Hilbert transform to compute phase locking value of two signals as a heuristic for brain synchrony

• Developed app in Typescript enabling platform-wide management of API settings, external-facing users, organizations, roles

• Created analytics dashboard to manage customer subscriptions and access API usage insights with consistent platform UI

• Improved usability, maintainability, scalability by developing custom React components, still in use across the platform

• Implemented AWS Lambda handler integrated with Python backend that improved consistency in MJML email templates and automated the email workflow with SES, saving costs and improving maintainability by deprecating Mailgun-based service