data science · machine learning · computer vision
Ahmed Mohammed
I build machine learning systems, computer vision tools, and full-stack data applications.
Incoming Information Sciences and Data Science student at the University of Illinois Urbana-Champaign, minoring in Computer Science. Based in Bartlett, Illinois.
// about
About
I'm finishing my Bachelor's in Information Sciences and Data Science at the University of Illinois Urbana-Champaign, with a minor in Computer Science. Before UIUC I earned my Associate of Science at Elgin Community College through dual credit courses in high school, working through C++, Python, and Calculus I to III before most people start college.
What draws me in is the point where an idea stops being theory and becomes something that actually runs. Machine learning, statistical modeling, and data systems all reward that mix of careful reasoning and real implementation, which is why I keep coming back to them. It's the same thread whether I'm building a computer vision tool that catches safety violations from an image, simulating how a species diverges over thousands of generations, or turning a piece of number theory into code that holds up at scale.
I like problems that are hard in an interesting way, and I care about doing the work well. Right now I'm focused on research and applied machine learning, and I'm always open to a good problem and good people.
// selected work
Projects
OSHA Vision Auditor
A computer vision tool that inspects a workplace for OSHA violations from images and video, no walkthrough required. It runs YOLOv8 for object detection and pairs it with Claude Vision to reason about what it sees, then generates a violation report with severity levels and remediation steps. I built it at a hackathon with a Next.js frontend, a FastAPI backend, and Supabase for persistence. The goal was simple: make it faster and easier for safety teams to stay compliant.
SEC Filings RAG
A retrieval-augmented generation system that lets you ask natural language questions about SEC 10-K filings from 10 major public companies. It pulls filings from EDGAR, chunks and embeds them into a vector database, retrieves relevant sections with semantic search, and generates grounded answers with citations. I built an evaluation suite that hit 100% retrieval accuracy across a 20-question test set. Try it out below - you get 3 questions per day.
Kalshi Signal Lab
A research pipeline that evaluates pricing efficiency across 408 resolved Kalshi prediction markets. It pulls market data from the Kalshi API, extracts market-implied probabilities from candlestick history at a 24-hour lead time before settlement, and measures calibration - Brier score 0.108, ECE 0.038 - on a held-out test set using a strict time-based 70/30 split. I fit an isotonic recalibration signal on the train set and backtested it out-of-sample across 57 bets with Kelly sizing and real Kalshi exchange fees. The signal didn't survive fees (-3.3% net ROI), which is the honest result. Three baselines validate the methodology and 47 unit tests cover every metric and the full backtest logic.
Maternal Mortality Analysis
A C++ data analysis tool that explores maternal mortality patterns from real-world CSV datasets. It supports full dataset inspection, single-factor breakdowns, pregnancy versus non-pregnancy condition reports, and multi-factor pattern exploration to surface correlations across demographics. I built it to practice working with raw tabular data in a low-level language without leaning on libraries - just the standard library, file I/O, and careful string parsing.
Speciation Simulation in Astyanax mexicanus
A stochastic Monte Carlo simulation in C++ that models how the Mexican cavefish splits into distinct populations over time. I modeled population size, mutation rate, fitness, and environmental pressure, then let it run. Cave populations diverged in roughly 2,800 generations while surface populations needed closer to 35,000, a gap that lines up with the biology. It's one of my favorite examples of a computational model capturing something genuinely complex.
// research
Research
DREAM Lab, UIUC
Undergraduate Research Assistant · advised by Professor Haohan Wang
2026 - present
I'm working on a recommendation reality gap detection system, which looks at where AI recommendations drift away from real-world outcomes. The idea is to mine organic reports from people on social platforms and use them as signal for when a recommendation did not hold up in practice.
Quest for Prime Numbers, UIC MURL Program
Undergraduate Researcher · advised by Professor Evangelos Kobotis
Jan 2026 - May 2026
Research on prime number algorithms through UIC's MURL program. The work was about taking big mathematical ideas and turning them into programs that actually run, then studying how different approaches perform once the numbers get very large. I liked it for the same reason I like most of my work: it sits right between careful reasoning and real implementation.
// tooling
Tech stack
Languages
ML & Data
Full Stack
Tools
// education
Education
University of Illinois Urbana-Champaign
2026 - 2028BS, Information Sciences + Data Science, minor in Computer Science
University of Illinois Chicago
2025 - 2026BS, Computer Science
Elgin Community College
Completed 2025Associate of Science, earned through dual credit coursework
Certification: Jira Fundamentals, Atlassian (2024)
// community
Beyond the code
Tutor, Kumon North America
2023 - 2026Tutored math and English at Kumon, helping students build fundamentals and the confidence to work through problems on their own.
Youth Leader, Al-Huda Youth Academy
2024 - 2025Led activities and mentorship for younger members of my community, focused on building a welcoming space where they could grow and take on responsibility.
// contact
Let's build something.
I'm open to research, internships, and collaboration in machine learning, computer vision, and data science. If you've got an interesting problem or just want to talk, email is the fastest way to reach me.
Status: open to opportunities