Hi, my name is

• Built an automated classification system for LLMOps case studies, fine-tuning ModernBERT to outperform commercial LLMs (e.g., Claude Haiku), demonstrating 30%+ accuracy, 69x faster inference, and 225× cost-efficiency.
• Developed a production-grade OCR pipeline supporting batch processing, smart caching, and artifact versioning for high-volume unstructured document workflows.
• Supported ZenML's marketing efforts by creating high-traffic technical content around MLOps, model evaluation, deployment, and workflow automation.

• Developed and maintained Vue.js and Ruby on Rails admin portal features, enabling merchants to manage digital ordering, loyalty programs, real-time order tracking, and promotional campaigns across multiple locations.
• Optimized merchant configuration interfaces through tabbed navigation and batched API requests—reducing API calls by 92% and significantly improving load times.
• Built an end-to-end global search system with fuzzy matching, weighted relevance scoring, full-text indexing, and keyword-based retrieval—enhancing users’ ability to quickly find and manage platform configurations.
• Strengthened role-based access control (RBAC) with CASL and CanCanCan, ensuring secure and synchronized authorization across the platform.

• Developed a VSCode extension (1500+ downloads) with a custom Python-based LSP server, integrating ZenML’s Python library into a TypeScript codebase to streamline MLOps workflows.
• Eliminated FastAPI server dependency by enabling direct library integration via users' local installations—simplifying version control and reducing setup overhead.
• Implemented real-time config monitoring using watchdog and an event-driven architecture for instant ZenML environment synchronization with the VSCode UI.
• Built a scalable AWS testing framework using LocalStack and Terraform, automating CI/CD pipelines with Docker.
• Authored comprehensive developer documentation, improving onboarding and maintainability for contributors.

• Built Pennant, a high-performance collaborative computational notebook, integrating a conflict-free replicated data type (CRDT) framework to enable real-time document synchronization with sub-50ms latency.
• Developed a distributed async execution engine using Flask, Celery, Redis, and RabbitMQ, optimizing request handling with background workers to achieve sub-100ms response times.
• Designed a React-based UI with TypeScript, React-Query, and Recoil, implementing browser and API caching that cut load times by 87.5%.
• Optimized infrastructure by reducing client-side TCP connections from linear to constant scale, cutting server resource consumption by 90% and improving system scalability.

• Featured projects:
  • CloudRounds – Permissions-Role-Based platform to manage access to educational seminars (React,TypeScript, Node.js, Prisma, GraphQL)
  • SnackBin – A database-backed application for receiving and debugging webhooks (Node.js, PostgreSQL, Socket.io)
  • Jabr – Discord-like chat and VoIP app (Next.js, Typescript, Prisma, MySQL, Planetscale, Socket.io)
  • SymptomOverflow – web app for analyzing symptoms and generating differential diagnoses (MERN stack)
  • CHF Manager – iOS app for the management of Congestive Heart Failure (Swift, SwiftUI)

Genesis Orthopedics & Sports Medicine

• Developed predictive models for total joint arthroplasty that identified high-risk patients with 87% accuracy, leading to an 18% reduction in clinic-wide readmission rates.
• Led research initiative analyzing cost burden reduction under BPCI/CJR models, resulting in $440K annual savings.