Malcolm / portfolio / 2026

Top 3 — Wayfair Hackathon

Identity

Computer Science student at the University of Illinois Urbana-Champaign with experience building AI document-processing systems, mobile products, computer vision tools, full-stack applications, and GPU-accelerated software.

AI systems

full-stack builds

computer vision

GPU programming

Software engineer / machine learning / systems

Malcolm
builds
intelligent systems

Software and machine learning engineer building scalable AI systems, consumer products, and high-performance applications. The site is organized like a project index: stripped back, direct, and built to let the work carry the weight.

About

Engineering depth with product instinct, shipped through real client work and systems-heavy builds.

Operating mode

I build software that moves between model logic, product surfaces, and system constraints without losing clarity. The work ranges from document intelligence and mobile apps to low-latency backends and CUDA implementations.

Current lens

Student at the University of Illinois Urbana-Champaign focused on AI products, full-stack execution, and machine learning initiatives.

Selected work

Project index

Experience

Built for teams and clients

Jun 2026 - Aug 2026

Wayfair

Software Engineering / Machine Learning Intern

Location

Boston, MA

Contributed to engineering and machine learning initiatives at Wayfair.

Built or supported systems related to data processing, experimentation, customer intelligence, or internal tooling.

Placed Top 3 in Wayfair's internal hackathon.

Mar 2026 - May 2026

Squint AI — CUBE Consulting

Software Engineer / AI Consultant

Location

Champaign, IL

Built an AI-powered pipeline that transformed PDF, DOCX, and TXT documents into structured standard operating procedures.

Developed structured JSON output for downstream editing and application use.

Created evaluation workflows using golden-file datasets.

Dec 2025 - Present

Nando Milano

Software Engineer

Location

Champaign, IL

Built a mobile restaurant loyalty application.

Developed customer rewards and points functionality.

Worked with Square Loyalty and Catalog APIs.

Jun 2024 - Aug 2025

Everyone Can Code Chicago

Software Engineering Intern / Technical Teaching Assistant

Location

Chicago, IL

Developed a computer-vision pipeline with 95% validation accuracy.

Integrated computer-vision functionality into an iOS application.

Supported more than 500 beta tests and App Store preparation.

Capabilities

A compact build language

Grouped by actual operating range instead of visual fluff. The goal is signal, not decoration.

Languages

Python

Languages

C++

Languages

CUDA

Languages

TypeScript

Languages

JavaScript

Languages

SQL

Languages

Swift

Languages

Java

Languages

Verilog

Languages

MIPS Assembly

AI and Data

Machine Learning

AI and Data

Computer Vision

AI and Data

Gemini API

AI and Data

OCR

AI and Data

Data Modeling

AI and Data

Document Parsing

AI and Data

Evaluation Pipelines

AI and Data

Pydantic

AI and Data

NumPy

Full-Stack

React

Full-Stack

React Native

Full-Stack

Next.js

Full-Stack

Expo

Full-Stack

FastAPI

Full-Stack

REST APIs

Full-Stack

Supabase

Full-Stack

Clerk

Full-Stack

Redis

Infrastructure and Tools

Git

Infrastructure and Tools

Docker

Infrastructure and Tools

AWS S3

Infrastructure and Tools

Celery

Infrastructure and Tools

Vercel

Infrastructure and Tools

Xcode

Infrastructure and Tools

Appium

Infrastructure and Tools

Tesseract

Infrastructure and Tools

Square APIs

Systems

GPU Programming

Systems

Parallel Computing

Systems

CUDA Kernels

Systems

Computer Architecture

Systems

Data Structures and Algorithms

Signals

Measured outcomes

Top 3

Wayfair internal hackathon finish

90%+

Parsing accuracy across 20+ SOPs for Squint AI

50K words

Approximately processed through the Squint AI pipeline

95%

Validation accuracy on an internship computer-vision pipeline

500+

Beta tests supported

<1 second

Word Hunt board solving time

Archive 01

DermaLens

Computer-vision application that improved classification accuracy by approximately 15% and provided mobile-friendly feedback.

PythonComputer VisionMachine Learning

Archive 02

Restaurant Event Website

Event website supporting more than 100 attendees, backed by normalized SQL data models and REST APIs.

JavaScriptSQLREST APIs

Archive 03

CUDA GPT-2

A systems-focused implementation of core GPT-2 forward-pass operations in CUDA, covering attention, matrix multiplication, normalization, residual connections, and GeLU.

CUDAC++GPU Programming

Archive 04

Mosaic Generator

A C++ image mosaic generator using KD-trees and nearest-neighbor matching.

C++Data StructuresKD-Trees

Archive 05

Maze Generator and Solver

A C++ application for generating and solving mazes using graph traversal and disjoint-set data structures.

C++Graph AlgorithmsData Structures

Archive 06

Huffman Compression

A lossless compression implementation using binary trees and Huffman encoding.

C++TreesCompression Algorithms

Contact

Open to engineering roles, selective freelance work, and AI product conversations.

If you want someone who can move from model behavior to product implementation to performance details, let's talk.