Practical software, built with curiosity.
I am a software developer working across Python, machine learning, and game development. I turn experiments, mechanics, and algorithmic ideas into focused tools people can actually use — built quickly, and built to last.
Detect and remove AI-writing patterns in lab reports — a transparent AI-likeness score with highlighted tells, plus a data-safe humanizer that rewrites prose without ever touching your numbers, units, or citations.
I am a developer focused on building practical software with Python. My work spans machine-learning experiments, game development, and small tools made to solve real problems quickly.
I enjoy turning ideas into working systems — combining creativity, automation, and intelligent problem-solving.
My goal is to keep building projects that are useful, well-made, and a little bit curious — shipped fast, then sharpened.
02 — Selected Work
A living archive of builds, pitches and research — pulled live from GitHub and curated by hand.
Detect and remove AI-writing patterns in lab reports — a transparent AI-likeness score with highlighted tells, plus a data-safe humanizer that rewrites prose without ever touching your numbers, units, or citations.
Make Claude a manager: Foreman routes your prompt to the best skills and agents, runs them in parallel like coworkers, keeps token spend low, and maintains a CLAUDE.md memory it updates as it works.
An ML classifier that detects prompt-injection attacks in LLM and agent inputs, benchmarked against real-world adversarial data from Gray Swan Arena.
A tiny smart compiler that watches how a language model "thinks" step by step, then builds faster shortcuts for it on the fly using LLVM compiler technology.
A full-stack application that helps students organize opportunities, deadlines, and applications in one place.
CodePulse is an AI-powered predictive code risk system built in Python. It analyzes GitHub repositories, studies commit history, and predicts which files are most likely to introduce bugs or instability before problems happen.
SafeWalk is a safety-focused navigation platform designed to help users choose safer walking routes using location intelligence and safety-aware routing systems. The project combines a modern interface with practical urban safety concepts to support smarter and safer movement.
DebtGuard is an AI-assisted app for debt decisions. Users input their finances to assess risk and simulate what-if scenarios, helping them understand outcomes and make better financial choices.
q-neo is a featured build with both its GitHub source and a Loom walkthrough, so it belongs in both the Project and Pitch views.
A detailed analysis of how Machine Learning and Deep Reinforcement Learning can be combined to shift from traditional market forecasting toward autonomous, sequential decision systems. The article explores state representation, model-based and model-free reinforcement learning approaches, risk-aware policy design, and adaptation in evolving market conditions.
AI Search Agent is a desktop chat application that launches from main.py and lets users ask web research questions in a simple GUI, processes them through its existing search-and-summarize pipeline with a background “Thinking…” state, and displays results directly in the chat window without saving any session data.
A pitch focused on increasing Loveable.ai users.
Calorie Tracker is a modern Python-based desktop app for tracking daily calorie intake, monitoring nutrition goals, and analyzing eating habits through an intuitive, app-like interface. It features structured data storage, daily summaries, and a clean dashboard experience designed for simplicity and performance.