Parkd is a camera-first parking enforcement platform that uses OCR-based license plate recognition (LPR) to give schools real-time visibility and control over their parking operations

This removes the need for physical permits entirely, and streamlines the parking process for admin, faculty, students and guests

I’m currently leading a team of 3 student engineers building Parkd from the ground up

A big part of my role has been turning this messy problem into something a small team can actually execute on.

I.e. defining the product vision, breaking complex challenges into milestones, and keeping engineering moving with clear priorities.

So I led our team as the architect behind a full multi-lot tracking system built on top of existing IP camera networks using RTSP stream ingestion. This involves:

• Working closely with school admin, campus safety, and IT to define the problem, and map out the desired workflow into an executable roadmap for engineers
• Working closely with engineers to scope features into testable builds, prioritize tradeoffs, and keep momentum high across an early-stage student team

So we came up with:

• License plate recognition pipeline to process live camera feeds with sub-3-second end-to-end latency, matching detected plates against a live permit database and triggering enforcement alerts in real time
• React + Vite admin dashboard backed by a Node/Express API layer, database infrastructure, and Redis caching for fast permit lookups and live system responsiveness
• Integrated Auth0 SSO to align with existing school identity systems, allowing secure admin access without disrupting current workflows
• Weekly engineering reviews, planning sessions, and brainstorming meetings to solve open-ended engineering problems such as entry/exit event modeling, lot occupancy tracking, duplicate detections, and partial plate reads
• CI/CD pipelines across dev, staging, and production using GitHub Actions and Docker, enabling faster releases, safer deployments, and isolated feature testing
• Admin and end-user flows in Figma, focusing on making guest pass purchases simple while giving enforcement staff clear, actionable live alerts

What I’ve enjoyed most about Parkd is that it sits at the intersection of engineering, product thinking, and leadership.

It’s not just about writing code, it’s about organizing people around a vision, making smart technical decisions, and building something that solves a real operational problem

Threaded Nexus is an always-on personal CRM and outreach copilot that transforms your everyday communication (emails, meetings, and interactions) into structured, actionable intelligence.

• It continuously ingests activity from tools like Gmail, LinkedIn messages, and Google Calendar
• From there, it builds a dynamic relationship graph of people, companies, and conversations, and uses LLMs to summarize threads, extract action items, and draft follow-ups

Instead of manually tracking relationships or letting opportunities slip, it proactively surfaces who to reach out to, what to say, and when

• All actions flow through a permissioned, auditable tool system, ensuring safe automation while maintaining human control.
• Built as a modern event-driven system with a robust data model and extensible architecture
• It replaces fragmented networking workflow with a single, intelligent layer for managing and scaling meaningful relationships

It's also a work in progress, more coming soon.

The Tower of Hanoi is a classic sort of child's game where you sort rings from one side of the toy onto the other without stacking a larger ring on a smaller one.

• Here I implemented this classic puzzle with an interactive web interface to demonstrate recursive algorithms
• I created a step-by-step visualization that shows the optimal solution path and allows users to experiment with different configurations

This project was to outline concepts I learned in CS data structures and algorithms courses to deepen my understanding of recursion and algorithm complexity

CrawlAI is a no-code platform for building AI assistants that combine web crawling, custom data, and LLMs into powerful, personalized workflows

• It's built by SCU students as a way to supercharge LLMs for research while using minimal tokens
• This is my fork of Ameen's original repository, built collaboratively with a team of SCU engineers to extend and enhance the LLM's capabilities
• We developed robust web scraping functionality and integrated advanced AI workflows to enable users to create sophisticated, data-driven automation without coding expertise.

I focused on the distribution and GTM for this tool

• When I joined the project, it was largely already built
• The team needed help in scaling the B2C tool and building feedback loops and user evals

This is my personal portfolio website from scratch using HTML, CSS, and JavaScript

• I've had this for a while now, and have been playing around with different UI styles, like dark mode functionality, and smooth navigation to create a modern, professional online presence

This project is just a space to showcase my work and experiences, what I've learned, and make a UI that sort of reflects my personality