"I don't just train models; I ship production AI systems."
I build machine learning systems that work in production.
Currently an ML/AI Engineer at Filly Coder, where I design and ship production FastAPI microservices powering a live EdTech marketplace, Filly Tutor. So far I have built three ML services from scratch to production: a TF-IDF + NLP tutor classification system (sub-1s latency, 95% precision), a multi-factor recommendation engine that replaced naive keyword matching, and a real-time DistilBERT sentiment analysis service with a 150–180ms average latency SLA and 85% test coverage on a 1,240-line codebase.
My background is Physics (First Class BSc, University of Ilorin), which means I approach ML problems with quantitative rigour, not just library calls. Before Filly Coder I spent a year as a Graduate Assistant at Fountain University, conducting applied ML research across healthcare (diabetes prediction, 78% accuracy on clinical data), agriculture (GPS-based plantation spatial analysis), and IoT business impact studies (214-respondent survey, chi-square testing, policy-grade reporting).
Beyond my day job, I build things I find interesting: ChaseBTC is an ensemble deep learning BTC trading system (LSTM + GRU + Conv1D, Sharpe Ratio 1.24, backtested with realistic fees and slippage) with three deployment interfaces. Little Botanist classifies 97 plant species using progressive EfficientNetB0 fine-tuning. MATRA is a multi-user AI thesis assistant deployed on HuggingFace Spaces, used by students to generate structured academic projects. Hey Rodea transcribes speech and scores it across six dimensions using Whisper ASR with int8 CPU optimization.
<aside> ⭐
Flagship Projects
Below are selected projects demonstrating my ability to take AI systems from idea to deployment.
</aside>
🔗 LinkedIn: https://www.linkedin.com/in/abdulrasheed-muhammed-abdulrasheed 💻 GitHub: themrandroid
📧 Email: abdulrasheedmuhammed002@gmail.com 📄 CV: Muhammed Abdulrasheed CV