Backend, Cloud & AI — building intelligent systems.
I build backend systems that run reliably in production—services, data pipelines, and the automation around them. I care about clear interfaces, low operational overhead, and designs that stay maintainable as requirements shift.
MS Computer Science at Stevens Institute of Technology. Previously built compliance microservices at Accenture. Focused on distributed systems and applied AI.
Selected projects across backend systems, applied AI, and full-stack builds.
Swipe to browse →
Navly
NJ Transit Intelligent Assistant
Personal
A full-stack navigation application for the NJ Transit system that combines high-performance spatial data processing with a RAG-based AI assistant. It transforms raw transit data into an interactive, conversational commuter experience.
Stack
Next.js · Django · PostgreSQL · OpenAI
Highlights(4)ShowHide
Architecting a specialized NJ Transit navigation engine utilizing GTFS data to simplify complex commuter schedules into a high-convenience, conversational interface.
Leading end-to-end development, including designing a relational schema for GTFS entities in PostgreSQL, building asynchronous data ingestion pipelines, and automating deployments via CI/CD.
Optimizing route discovery and spatial lookups by implementing PostGIS spatial indexing, reducing query latency for dense transit datasets.
Engineering a RAG-powered AI assistant that grounds LLM responses in verified transit schedules, enabling users to query complex NJ Transit data through natural language instead of manual timetable filtering.
An intelligent safety net that understands the context of your environment and acts on your behalf when you cannot, using real-time audio transcription and risk assessment.
A real-time financial planner that anticipates spending using proactive AI agents. Ledgerly provides actionable insights to help users make smarter budgeting decisions before they swipe their cards.
Built a platform designed to take the guesswork out of personal finance by integrating a comprehensive tech stack in 24 hours at TartanHacks 26, Carnegie Mellon University.
Developed a real-time financial planner that anticipates spending and provides actionable insights.
Integrated Plaid for real-time financial statements and Ripple for payments.
Utilized Agentuity to power proactive AI agents for smarter budgeting.
An AI-powered discovery tool that suggests the perfect "Magic Middle" meeting point for friends—whether they are coming from different boroughs, suburbs, or neighboring cities. It removes the friction of decision-making by finding fair-distance, vibe-matched venues.
Stack
Lovable · Phinite · Tavily · AI Agents
Highlights(4)ShowHide
Built an AI-powered discovery tool that suggests the perfect "Magic Middle" meeting point for friends in a 2-hour hackathon.
Streamlined the mobile-first UI and managed the cloud database using Lovable.
Orchestrated the "AI Brain" with Phinite to logic-map transit midpoints and user preferences.
Utilized Tavily for real-time search intelligence to discover local community gems.
A full-stack navigation application for the NJ Transit system that combines high-performance spatial data processing with a RAG-based AI assistant. It transforms raw transit data into an interactive, conversational commuter experience.
Stack
Next.js · Django · PostgreSQL · OpenAI
Highlights(4)ShowHide
Architecting a specialized NJ Transit navigation engine utilizing GTFS data to simplify complex commuter schedules into a high-convenience, conversational interface.
Leading end-to-end development, including designing a relational schema for GTFS entities in PostgreSQL, building asynchronous data ingestion pipelines, and automating deployments via CI/CD.
Optimizing route discovery and spatial lookups by implementing PostGIS spatial indexing, reducing query latency for dense transit datasets.
Engineering a RAG-powered AI assistant that grounds LLM responses in verified transit schedules, enabling users to query complex NJ Transit data through natural language instead of manual timetable filtering.
An intelligent safety net that understands the context of your environment and acts on your behalf when you cannot, using real-time audio transcription and risk assessment.
A real-time financial planner that anticipates spending using proactive AI agents. Ledgerly provides actionable insights to help users make smarter budgeting decisions before they swipe their cards.
Built a platform designed to take the guesswork out of personal finance by integrating a comprehensive tech stack in 24 hours at TartanHacks 26, Carnegie Mellon University.
Developed a real-time financial planner that anticipates spending and provides actionable insights.
Integrated Plaid for real-time financial statements and Ripple for payments.
Utilized Agentuity to power proactive AI agents for smarter budgeting.
An AI-powered discovery tool that suggests the perfect "Magic Middle" meeting point for friends—whether they are coming from different boroughs, suburbs, or neighboring cities. It removes the friction of decision-making by finding fair-distance, vibe-matched venues.
Stack
Lovable · Phinite · Tavily · AI Agents
Highlights(4)ShowHide
Built an AI-powered discovery tool that suggests the perfect "Magic Middle" meeting point for friends in a 2-hour hackathon.
Streamlined the mobile-first UI and managed the cloud database using Lovable.
Orchestrated the "AI Brain" with Phinite to logic-map transit midpoints and user preferences.
Utilized Tavily for real-time search intelligence to discover local community gems.
Backend engineer on trading compliance microservices for a US asset management client.
Engineered and scaled trading compliance microservices in Java (Spring Boot) on AWS for a US-based asset management client, exposing pre-trade and post-trade compliance APIs that supported 1,000+ daily trades.
Designed and shipped a scalable, Python-based internal reporting tool for portfolio managers that replaced manual checks, reducing guideline-monitoring review time by over 90%.
Collaborated on a RAG-based GenAI prototype over PDF research reports, reducing manual analysis efforts by over 95%.
Developed Python scripts to automate test case generation for new backend features, increasing release confidence for each deployment and cutting manual QA effort by 90%.
Built a Java-based API automation and data reconciliation framework (JSON/Excel), reducing manual regression testing effort by 70%.
Upgraded legacy services from Java 8 to 17 and refactored CloudFormation IaC, enabling compliant, zero-downtime deployments and cleaning up long-standing compatibility issues for 10+ microservices.
Led the migration of test automation services from a decommissioned AWS region, updating environments, Jenkins pipelines, and Linux configs to cut infrastructure waste while keeping regression runs uninterrupted for 10+ microservices.
Ideas
Concepts and early-stage products I am exploring outside of client work.
Swipe to browse →
Cachd
Capture now. Find it later.
Early concept / pre-MVP
Cachd saves screenshots, links, and posts to your private workspace when you share them from your phone—then lets you browse or search everything in one clean place on the web.
Highlights(4)ShowHide
Share screenshots, links, and posts from Instagram, Safari, X, and more via an iOS Share Extension.
Everything lands in a private, searchable workspace—no folders or manual tagging in v1.
Stop losing inspiration trapped in your camera roll or scattered across apps.
Building the first version now—early access on iOS and web.
A digital wardrobe experience that helps users organize what they own, plan outfits, and prepare for future occasions—making wardrobe management more intentional.
Highlights(4)ShowHide
Organize what you own and plan outfits in one place.
Prepare for future occasions with occasion-based planning.
Detect gaps in your wardrobe based on what you already have.
Smarter purchase recommendations grounded in your existing closet—not generic suggestions.
Transform urban living with subscription-based agriculture plots. Manage your own farmland remotely through an app, grow organic vegetables, visit on weekends to harvest—or get them delivered.
Highlights(4)ShowHide
Subscription-based agriculture plots designed for urban living.
Manage farmland remotely through a mobile app.
Grow organic vegetables and visit on weekends to harvest—or get them delivered.
Unused yield turns into credits redeemable with partners or on future subscriptions.
Cachd saves screenshots, links, and posts to your private workspace when you share them from your phone—then lets you browse or search everything in one clean place on the web.
Highlights(4)ShowHide
Share screenshots, links, and posts from Instagram, Safari, X, and more via an iOS Share Extension.
Everything lands in a private, searchable workspace—no folders or manual tagging in v1.
Stop losing inspiration trapped in your camera roll or scattered across apps.
Building the first version now—early access on iOS and web.
A digital wardrobe experience that helps users organize what they own, plan outfits, and prepare for future occasions—making wardrobe management more intentional.
Highlights(4)ShowHide
Organize what you own and plan outfits in one place.
Prepare for future occasions with occasion-based planning.
Detect gaps in your wardrobe based on what you already have.
Smarter purchase recommendations grounded in your existing closet—not generic suggestions.
Transform urban living with subscription-based agriculture plots. Manage your own farmland remotely through an app, grow organic vegetables, visit on weekends to harvest—or get them delivered.
Highlights(4)ShowHide
Subscription-based agriculture plots designed for urban living.
Manage farmland remotely through a mobile app.
Grow organic vegetables and visit on weekends to harvest—or get them delivered.
Unused yield turns into credits redeemable with partners or on future subscriptions.