Minh's Portfolio

Research Project

cGANs for Candidate Pseudo-Absence Generation

Conditional GANs for Candidate Pseudo-Absence Generation in Crop Suitability Modeling

Project Description: This research project explores the application of conditional GANs (cGANs) for generating candidate pseudo-absences in crop suitability modeling. The goal is to create synthetic data that can be used to improve the accuracy and robustness of crop suitability models.

Designing and implementing a Conditional GAN (cGAN) pipeline in PyTorch to generate synthetic pseudo-absence data for crop suitability modeling, addressing class imbalance in presence-only environmental datasets. Building end-to-end geospatial ML pipelines in Python using NumPy, Pandas, and raster-processing workflows to extract, transform, and analyze high-dimensional satellite and bioclimatic data. Engineering and evaluating feature-generation workflows using statistical analysis and PCA-based distribution comparison to improve training stability and guide hyperparameter tuning of generative models.

Research Poster

Web App

Grounded Synthetic Respondent Lab

Grounded Synthetic Respondent Lab - AI Persona & Survey Simulation Platform

Web Link

Tech Stack: Next.js, TypeScript, FastAPI, PostgreSQL, Clerk, Docker

Built a full-stack application for an AI-powered research workflow that creates grounded synthetic personas, runs model-backed survey simulations, and generates analysis dashboards, reliability checks, interview synthesis, and LLM-summarized insights. Implemented end-to-end product flows across Next.js, TypeScript, Tailwind, Framer Motion, FastAPI, SQLAlchemy, Alembic, and PostgreSQL, including audience/product/market setup, survey ingestion, persona preview, simulation runs, result dashboards, and insights sections. Added production-grade deployment and security hardening with Clerk invite-only auth, ownership enforcement, trusted backend proxying, SSRF protection, upload guardrails, daily per-user quotas, in-flight provider-run locks, Docker startup migrations, Render backend hosting, Neon Postgres, and Vercel frontend deployment.

Source Code

Web App

BackTrack

BackTrack - Learning Recovery Web App

Web Link

Tech Stack: Next.js, TypeScript, FastAPI, Python, LLM APIs

Backtrack is a prerequisite-aware learning recovery web app. It diagnoses why a student is stuck and builds a personalized path to help them catch up. Students can upload course materials such as a syllabus or lecture slides. Backtrack then extracts the main course topics, identifies likely missing prerequisite concepts, explains why those prerequisites matter, recommends free learning resources, generates a short readiness quiz, and provides a focused tutor chat. Backtrack is not a generic AI tutor. It is a course-aware learning recovery system.

Source Code

Web App

CyberCoach

CyberCoach - Scam Detection & Safety Assistant

Web Link

Tech Stack: Next.js, TypeScript, FastAPI, Python, OpenRouter, Pytest, Playwright

A full-stack scam-detection platform across message, URL, screenshot, document, and call-analysis workflows, combining rule-based heuristics, OCR/document parsing, URL reputation and destination checks, and dual-model AI review into plain-language risk results. Designed a FastAPI analysis backend with unified scan-response adapters, privacy-mode redaction, report export, upload validation, rate limiting, and SSRF-hardened URL inspection to safely handle user-submitted messages, links, files, and recordings. Added automated test coverage with pytest, Vitest, and Playwright for scan APIs, privacy/history regressions, oversized uploads, SSRF cases, report escaping, and demo fixtures for repeatable AI decision-trace outcomes.

Source Code

Web App

SlotPilot

SlotPilot - Booking System

Web Link

Tech Stack: Next.js, TypeScript, Supabase, PostgreSQL, Tailwind CSS, Radix UI, Playwright

Developed a full-stack appointment booking and waitlist recovery platform for local businesses with public booking pages, authenticated dashboards, onboarding, service management, and scheduling tools. Implemented booking lifecycle, availability, rescheduling, waitlist offers, cancellation recovery, Supabase Auth, PostgreSQL RLS policies, and transactional email notifications. Built responsive dashboard and customer-facing UI with React, Tailwind CSS, Radix UI, Framer Motion, Recharts, React Hook Form, and Zod, with coverage from Vitest and Playwright tests.

Source Code

Mobile App

ShotFrame

App Link

Tech Stack: Flutter, Dart, Riverpod, GoRouter, Supabase, PostgreSQL, Deno, Drift/SQLite

Built a Flutter mobile app for photographers to search references, save boards, create shot lists, and run offline shoot sessions with local-first Drift/SQLite persistence. Integrated Supabase Auth, Postgres, and Deno Edge Functions for secure backend workflows including Unsplash search proxying, account deletion, rate limiting, and guest-to-account data migration.

Source Code

Web App

Aperture AI

Aperture AI - AI-Powered Image Editor

Web Link

Tech Stack: React, TypeScript, Node.js, Express, OpenAI DALL-E, Tailwind CSS, PostgreSQL, Vite

Developed a full-stack AI image editing platform that enables users to transform and enhance photos using natural language prompts. Implemented AI-powered image generation and editing with OpenAI's DALL-E, and built a responsive frontend with React and Tailwind CSS. The application features real-time previews, edit history, and a user-friendly interface for seamless image manipulation. Deployed with performance optimizations for efficient AI processing on cloud infrastructure.

Source Code

Mobile App

Perfect Form

Tech Stack: Flutter, Firebase, Python, Flask, MediaPipe, OpenCV (CV2)

Developed a cross-platform fitness app using Flutter that allows users to upload workout photos and receive real-time feedback on their form. Implemented pose detection using MediaPipe and OpenCV (CV2) on a Python Flask backend, and stored user data and pose comparisons with Firebase. The system calculates joint angles and highlights misalignments to help users improve technique and reduce injury risk.

Source Code

Live Demo

Web App

CS - Catalog

Tech Stack: React, Bootstrap, Node.js, Express, PostgreSQL, Railway, JWT

Collaborated with a team to build a full-stack web app that helps users log and track their progress through machine learning topics. Developed a Node.js/Express backend with a PostgreSQL database to handle user data and topic completion tracking. Integrated JWT-based authentication for secure login and access. Designed a responsive and user-friendly frontend using React and Bootstrap. Deployment is in progress as we prepare the app for production on Railway.

Source Code

Live Demo

Chatbot

Behavioral Interview

Behavioral Interview Chatbot

Tech Stack: Flask, HTML, CSS, Bootstrap, JavaScript, Gemini 1.5 API

Built an AI-powered customized knowledge chatbot using Python, Flask, and the Gemini 1.5 API to help users practice behavioral interview questions and receive real-time, AI-generated feedback. Designed and implemented a clean, responsive frontend using HTML, CSS, Bootstrap, and JavaScript. Secured API integration using environment variable management to protect user data and improve reliability.

Source Code

Semester Project

Face Mask Detection

Tech Stack: Python, TensorFlow, Scikit-learn, Keras, OpenCV, Matplotlib

Developed a deep learning model for three-class face mask detection, achieving 99.31% accuracy and a strong F1-score on a dataset of over 15,000 images. We engineered a robust data pipeline with preprocessing and augmentation to improve generalization, and designed a custom Convolutional Neural Network (CNN) in Keras/TensorFlow. Model training was managed in Google Colab with GPU acceleration and early stopping to prevent overfitting, while performance was visualized through accuracy/loss curves and confusion matrices. The best-trained model was optimized and saved for real-time tracking applications.

Source Code

Creative

Photography

Creative

Hello, I'm

Minh Nhat Doan

About Me

I am Minh Nhat Doan

My Resume

AA Computer Science

BS Computer Science and Data Science minor

Frontend Developer and Ad Ops Intern

Photographer

Software Development and Data Management Co-op

Undergraduate Research Scholar

Python

Java

C#

SQL

JavaScript

Git

Postman

Visual Studio

Adobe Suite

Figma

HTML

CSS

Next.js

React.js

Flutter

RESTful APIs

ASP.NET Core

Flask

PostgreSQL

Node.js

Express.js

Portfolio

Contact Me

Contact Info

nhatdoanminh2303@gmail.com

Social Info

Send Message