Firebase Firestore for persistence, with a static data module as the current source of truth for 300+ test records.

Clinical metadata engine classifies each result into zones using CLL-specific thresholds and generates physician-level context.

React 19 components for blood test cards, visual gauges, trend charts, and urgent findings -- all built with Tailwind CSS.

Recharts powers interactive line charts with reference zones, brush zoom, and animated tooltips for temporal analysis.

Mobile-first, Offline-capable, Accessible

A patient-focused PWA that transforms raw blood test data into clinically-contextualized, interactive visualizations for tracking chronic lymphocytic leukemia (CLL) treatment response.

Each test is evaluated against CLL-specific thresholds -- not generic lab ranges. Zones map to Critical, Borderline, and Normal with physician-reviewed messages.

Clinical Zone Engine

Recharts-powered line charts with color-coded background zones, reference lines, and a brush control for zooming into any time period across 5+ years of data.

Interactive Trend Charts

Horizontal gauge bar with dynamic scaling, threshold markers, and diagonal stripe patterns in critical zones for color-blind accessibility.

Visual Gauge

Critical conditions like severe pancytopenia, acute kidney injury, and hepatitis are surfaced in a prominent alert banner at the top of the dashboard.

Urgent Findings

Installable as a Progressive Web App with offline capability -- essential for hospital and clinic environments where WiFi is unreliable.

PWA / Offline Access

WCAG-compliant with ARIA labels, semantic HTML, keyboard navigation, focus-visible rings, and color-blind safe patterns throughout.

Accessibility-First

A family member undergoing treatment for Chronic Lymphocytic Leukemia (CLL) was receiving regular blood test reports -- dense PDF tables with dozens of values but no narrative. Understanding which numbers were improving, which were concerning, and how they trended over years of treatment required a physician to interpret. Between appointments, the patient and family had no way to quickly assess the trajectory of their fight.

Raw lab reports lack clinical context for patients

Rather than building a generic health dashboard, the app was designed around one patient's specific treatment journey. A clinical metadata engine maps each test result against CLL-specific thresholds with physician-reviewed zone definitions. Interactive gauges and historical trend charts let the family see at a glance whether values are improving, stable, or concerning.

Build a CLL-specific, clinician-informed visualization layer

The family can now open the app on any device and within seconds understand the status of all 12 tracked biomarkers. Historical trend charts spanning 5+ years of treatment data tell the story of recovery at a glance. Urgent findings are surfaced prominently so nothing is missed. The app serves as a bridge between appointments -- translating clinical data into a language the whole family can read.

Immediate clarity and peace of mind

Blood Test Visualizer

Open-Meteo API provides 80+ years of historical weather data and forecasts for any location worldwide.

Progressive data loading fetches 10 years at a time, while the comparison engine aligns historical data by calendar date.

React 19 components provide three weather comparison views plus astronomical tools including moon phases and golden hour.

Service workers cache the entire app shell via Workbox for offline-first reliability with automatic update detection.

Web, Installable PWA, Native Android via Capacitor

A progressive web app that visualizes historical weather data for any location worldwide, featuring interactive astronomical tools, multi-year comparison views, and offline-first architecture.

Select any date and instantly compare that day's weather across every available year back to 1940.

Same-Day Comparison

Interactive slider-driven tool rendering accurate moon phases using SVG with crater details and gradient shading.

Moon Phase Visualizer

Draggable, rotatable Earth visualization rendered from NASA Blue Marble satellite imagery.

Interactive Earth Globe

Sunrise, sunset, and photographer-friendly golden hour and blue hour windows for any location.

Golden Hour Calculator

Service workers cache the entire app shell for instant loading without a network connection.

Offline-First PWA

Single React codebase serves web, installable PWA, and native Android via Capacitor.

Cross-Platform

Existing weather apps focus almost exclusively on forecasts -- what is coming next. But understanding climate means looking backward. There was no lightweight, mobile-friendly tool that let a casual user visually compare today's weather to the same day 10, 20, or 50 years ago without spreadsheets or API keys.

Weather data is abundant but rarely tells a story

Rather than tacking historical data onto a forecast app, the solution was built from the ground up around comparison and context. A compare-first architecture shows the same calendar day across every available year, making trends immediately visible without constructing queries.

Treat historical weather as a first-class experience

A production-ready application that transforms raw meteorological archives into an intuitive visual experience. Users can explore historical weather for any city worldwide and interact with astronomical tools that add meaning to the data.

A polished tool for exploring 80+ years of weather

Weather History

Historical listing data sourced from Inside AirBnB, providing neighborhood-level market intelligence.

Automated EDA pipeline handles data cleaning, feature engineering, and generates HTML reports for data quality review.

Linear regression model trained on neighborhood, property type, amenities, and room count to predict listing profitability.

Dockerized Flask backend serves prediction results and market data through a clean REST API.

Interactive Map, Profitability Metrics, Seasonal Charts

A map-based dashboard application for AirBnB investors that uses linear regression to predict how profitable a future listing might be based on historical data.

Linear regression forecasts how profitable a potential listing could be based on neighborhood, property type, amenities, and room count.

Profitability Prediction

Explore AirBnB listings geographically with profitability data, average daily rates, and occupancy metrics.

Interactive Map Dashboard

Projects how seasonality affects revenue throughout the year, showing peak and off-peak periods.

Seasonal Impact Analysis

Estimates annual revenue, average daily rate, and occupancy rate projections for prospective listings.

Revenue Forecasting

Displays how currently active listings in a given area are performing, providing real-time investment benchmarks.

Market Benchmarking

Raw listing data is preprocessed through EDA with automated HTML report generation for data quality review.

Automated Data Pipeline

The AirBnB market is highly competitive and unpredictable. Investors considering rental property purchases need reliable data to make informed decisions, but there is no simple, unified tool that visualizes market conditions, estimates potential revenue, and accounts for seasonal trends in one place.

Investors lack data-driven tools for AirBnB decisions

The solution aggregates AirBnB listing data from Inside AirBnB and applies linear regression models to predict key profitability metrics. Results are presented through an interactive map interface where investors can explore listings by location and compare performance data visually.

Build a map-based predictive analytics dashboard

A full-stack application that gives AirBnB investors a data-driven edge. The dashboard provides profitability predictions, seasonal revenue projections, and real-time market benchmarks -- all through an intuitive map-based interface.

A comprehensive investment analysis tool

AirBnB Listing Analyzer

Real-time market data feeds, Shopify Storefront API for merchandise, and Stripe for secure payment processing.

GraphQL provides a flexible data layer for efficiently querying market data, content, and commerce operations.

Next.js handles server-side rendering for fast initial loads and SEO, with API routes for backend logic.

Responsive React components for the analytics dashboard, Stripe checkout flow, and Shopify-powered storefront.

A modern cryptocurrency platform built with Next.js featuring real-time market data, integrated payment processing via Stripe, and a Shopify-powered merchandise storefront.

Live market data and analytics displayed through an intuitive control panel with responsive charts and key metrics.

Real-Time Dashboard

PCI-compliant payment processing integrated seamlessly for secure transactions and invoice management.

Stripe Payments

Branded merchandise catalog powered by Shopify Storefront API with integrated cart and checkout flow.

Shopify Storefront

Flexible, efficient data fetching via GraphQL for market data, content, and commerce operations.

GraphQL Data Layer

Next.js server-side rendering delivers fast initial loads, improved SEO, and optimal Core Web Vitals.

SSR Performance

Mobile-first layout ensures a consistent experience across desktop, tablet, and smartphone screens.

Responsive Design

Decentralized finance projects often lack polished web experiences that inspire user confidence. DeFi X needed a modern platform that could display real-time market data, process payments securely, and sell branded merchandise -- all while maintaining the fast, responsive experience crypto-native users expect.

DeFi projects need professional, trust-building web presence

Leveraged Next.js for server-side rendering and optimal performance, integrated Stripe for secure payment processing, and connected a Shopify storefront for merchandise sales. GraphQL provided a flexible data layer for querying market data and managing content.

Build a performant Next.js platform with integrated commerce

A fully functional cryptocurrency website with integrated payment processing and merchandise sales, delivered on an aggressive 3-week timeline. The platform provides a polished, trustworthy experience that serves both informational and commercial needs.

A professional crypto platform delivered in 3 weeks

Crypto Website Development for DeFi X

Interactive controls let users select sorting algorithms, adjust animation speed, and configure array sizes for different experiments.

Five classic sorting algorithms implemented with async/await patterns that yield control between operations for smooth animation.

Async controller manages step-by-step execution, color-codes comparisons and swaps, and maintains animation state.

Real-time animated bar chart, responsive across all devices

An interactive real-time animated algorithm visualizer built with React.js that brings sorting algorithms to life through dynamic visual representations.

Supports Bubble Sort, Selection Sort, Insertion Sort, Merge Sort, and Quick Sort for side-by-side comparison.

Multiple Algorithms

Control animation speed and array size to observe step-by-step execution or large-scale performance differences.

Adjustable Speed & Size

Color-coded operations -- comparisons, swaps, and placements -- animated in real time with immediate visual feedback.

Real-Time Animation

Clean CSS with responsive layouts ensuring smooth operation across different screen sizes and devices.

Sorting algorithms are a foundational concept in computer science, yet understanding how they work purely through code or pseudocode can be abstract and difficult to grasp. Students and developers struggle to differentiate between the behavior and efficiency of algorithms without a visual frame of reference.

Sorting algorithms are abstract without visualization

The solution was a fully interactive web application that renders sorting algorithms as animated bar charts in real time. Each algorithmic operation is color-coded and animated, providing immediate visual feedback on how the algorithm traverses and manipulates data.

Build an interactive, animated sorting experience

A polished, interactive application that makes abstract computer science concepts tangible. Users can visually observe how different sorting algorithms process data, compare their behavior side by side, and build genuine intuition for algorithmic efficiency.

An intuitive educational tool for algorithm learning

Sorting Algorithm Visualizer

Low-resolution photometric measurements from astronomical surveys, targeting cost-effective classification without spectroscopic imaging.

Comprehensive preprocessing pipeline handles missing values, normalizes features, and balances class distributions before training.

Three classical machine learning algorithms trained and tuned in separate Jupyter notebooks with consistent evaluation metrics.

Advanced models including gradient boosting, probabilistic clustering, and deep learning to capture complex nonlinear relationships.

Systematic evaluation using accuracy, precision, recall, and F1-score for objective algorithm comparison across all models.

A machine learning research project that reduces imaging costs by leveraging cost-effective low-resolution photometric measurements to accurately classify astronomical point sources.

Six classification algorithms tested and benchmarked -- KNN, Decision Tree, SVM, XGBoost, Gaussian Mixture, and Neural Network.

Multi-Model Pipeline

Uses low-resolution photometric data instead of expensive high-resolution spectroscopic imaging for celestial classification.

Cost Reduction

Consistent metrics including accuracy, precision, recall, and F1-score across all models for objective comparison.

Systematic Evaluation

Separate Jupyter notebooks for each algorithm with clear documentation and reproducible workflows.

Reproducible Research

Neural network implemented in Python to capture complex nonlinear relationships between photometric features and source classifications.

Deep Learning

Comprehensive preprocessing handling missing values, feature scaling, and class balancing before model training.

Data Pipeline

Traditional classification of astronomical point sources -- stars, quasars, and galaxies -- relies heavily on high-resolution spectroscopic imaging, which is both time-consuming and expensive. As the volume of astronomical survey data grows exponentially, there is a pressing need for automated methods that work with cost-effective, low-resolution photometric data.

Spectroscopic imaging is too expensive for large-scale classification

The project followed a structured data science workflow, implementing six distinct classification algorithms and evaluating each against consistent metrics. Every model was trained, tuned, and benchmarked in separate Jupyter notebooks with clear documentation for full reproducibility.

Build and benchmark a multi-model classification pipeline

A complete, reproducible ML pipeline that demonstrates photometric data can replace expensive spectroscopic imaging for celestial object classification. The systematic comparison across six algorithms provides clear evidence for which approaches work best in this domain.

Cost-effective classification with systematic model comparison

Classification of Point Sources in Space

JC+ source files containing class definitions with inheritance, methods, and bounds-checked arrays.

Custom tokenizer breaks JC+ source into typed tokens including keywords, identifiers, operators, and literals.

Recursive descent parser builds an Abstract Syntax Tree from the token stream, enforcing grammar rules.

Type system validates subtyping relationships, method signatures, and inheritance hierarchies for correctness.

Translates the typed AST into valid C code, injecting runtime bounds checks for array access safety.

Compiled C programs via standard toolchains

A custom programming language compiler written in Java that adds class-based inheritance, subtyping, and bounds-checked arrays to the C programming language.

Define classes with instance variables, constructors, and methods that extend other classes using Java-style inheritance.

Class-Based Inheritance

Full subtyping allowing subclass instances wherever a superclass is expected, enabling polymorphic behavior.

Subtyping System

Runtime bounds checking prevents out-of-bounds reads and writes -- addressing C's most dangerous security vulnerability.

Bounds-Checked Arrays

Traditional pipeline with lexer, parser, typechecker, and code generator -- each as a distinct module.

Multi-Stage Compiler

Integer types, void, array types, arithmetic and comparison operators, method calls, and object instantiation.

Built-In Types & Expressions

Full support for if/else conditionals, while loops with break statements, return statements, and block-scoped grouping.

Control Flow

C is a powerful low-level language, but it lacks modern features like object-oriented programming, type-safe inheritance, and automatic bounds checking for arrays -- the latter being a well-known source of security vulnerabilities. Developers need these higher-level constructs without sacrificing C's performance and portability.

C lacks modern safety and object-oriented features

JC+ was designed to bring higher-level programming constructs to a C-like language while compiling down to standard C. The compiler follows a traditional multi-stage pipeline -- lexer, parser, typechecker, and code generator -- each implemented as a distinct Java module.

Design a new language with a multi-stage compiler targeting C

A complete, functional compiler that translates JC+ source code into valid C programs. The language combines the expressiveness of object-oriented languages with C's performance, proving that high-level features can compile down cleanly to procedural constructs.

Open-Source & Repositories