Mkareos
Kzkoz

B.Computing · UofGUniversity of Guelph · Open to internships
Mkareos Kzkoz
About

I'm a second-year Software Engineering student at the University of Guelph. I build across the whole stack, from C and OpenSSL down low to React and TypeScript up top, and I care most about the parts people tend to gloss over: retrieval quality, real benchmark numbers, tamper tests, the failure cases.

Recently that's looked like an explainable RAG search engine, a post-quantum cryptography lab, and a deterministic scoring engine that took 2nd at GDGHacks. I'm open to software internships where I can work on hard problems alongside people who've been at it longer.

Now
building Lattix, a post-quantum cryptography migration lab
studying 2nd year software engineering
won 2nd place · GDGHacks 2026 · Best Cutting Edge UI category
Projects
Verivance Explainable RAG search engine. Chunks and embeds documents into Pinecone (281-chunk demo corpus), pulls the top 5 semantic matches per query, ranks them, and passes them to Gemini as grounding. Every answer ships with source citations and similarity scores. FastAPI backend, Next.js frontend on Vercel.
Next.jsFastAPIPinecone
2026 GitHub ↗
Lattix Interactive lab for comparing classical and post-quantum cryptography side by side. C/OpenSSL backend wrapping ML-KEM-768, ML-DSA-65, and AES-256-GCM (key generation, encapsulation/decapsulation, sign/verify, authenticated encryption) with timing benchmarks and tamper tests, exposed to a React frontend through a Node/TypeScript API. Built the C crypto layer and the API bridge.
COpenSSLML-KEM
2026 GitHub ↗
RankedIn 2nd · GDGHacks Turns LinkedIn profiles into an explainable 0–2500 "Aura Score" across 8 weighted dimensions on a live public leaderboard. Custom deterministic scoring engine (no AI), plus a 3-tier scraping fallback (fresh scraper, enrichment API, raw HTML parsing) that keeps data flowing past anti-scraping defenses.
Next.jsTypeScriptSupabase
2026 GitHub ↗
Visual Mouse Hands-free cursor control from a webcam. MediaPipe hand-landmark detection and OpenCV frame processing drive OS-level mouse actions through PyAutoGUI, with click and move logic derived from relative distances between finger landmarks. Gesture recognition is kept separate from cursor control so the interaction stays responsive and tunable.
PythonOpenCVMediaPipe
2026
Panic Pick Responsive pixel-art browser game in vanilla JavaScript, HTML, and CSS. Animated day/night themes, custom UI transitions, crisp pixel rendering, and interactive menus, with the UI and game logic split into reusable components so features drop in without a framework or build step.
HTMLCSSJS
2026 GitHub ↗
Stack
languages Python, C, Java, JavaScript, TypeScript, SQL, Bash, MATLAB, HTML, CSS
frameworks React, Next.js, Node.js, Express, FastAPI, Tailwind, Framer Motion, PyTorch, Pandas, OpenCV, MediaPipe, PyAutoGUI
tools Git, GitHub, CMake, OpenSSL, Vite, Linux, Vercel, VS Code, npm/pip, LaTeX, Figma, Pinecone, PostgreSQL, Supabase
Contact

Open to internships, collabs, or just a good conversation. Reach out.