Software & ML Engineer (Student) | Python • Rust • React • Data Engineering
I love building systems that are fast, intelligent, and actually usefulm from high-performance search engines in Rust to ML models that explain their predictions. Passionate about AI/ML, full-stack engineering, system design, automation, and everything F1.
Predicting top-3 F1 race outcomes using Gradient Boosting Regressor + FastF1 Telemetry + Weather Data.
Also experimenting with Gemma 3.0 for natural-language race insights (e.g., “Why did Mercedes dominate 2014–2020?”).
A lightning-fast local search engine using:
- Tantivy for full-text indexing
- ANN-based embedding store for semantic search
- Sub-millisecond lookup
- ~1GB vector store indexing ~100GB of files
- Multi-threaded preprocessing + memory mapping
Offline transit system that:
- Loads GTFS schedule + real-time data
- Renders 500+ route points
- Shows live positions, next stations, trajectory playback
- 80% faster data recomputation using optimized caching
Hybrid CNN-ViT model for severity grading (“Mild/Moderate/Severe”).
Includes full preprocessing, proxy labeling, Grad-CAM explanations.
Under review at AIHC 2025.
Local credential manager with encryption, salting, hashing, and secure storage.
Custom document Q&A system using vector embeddings, FAISS, and LLMs.
CNN for 10-class image recognition with augmentation + visualization.
Python • Rust • JavaScript (ES6+) • Java • C • C++ • C# (learning) • SQL
HTML • CSS • React.js • Tailwind (experience) • Basic Angular
Node.js • Express.js • FastAPI • REST APIs
PyTorch • TensorFlow • Scikit-learn • XGBoost • Transformers • LangChain
Data Processing • Feature Engineering • Computer Vision • NLP
MySQL • SQL Server • MongoDB • SQL.js • Tantivy • ANN Vector Search
Docker (learning) • Linux (learning) • Git • VS Code
FastF1 • Gemini API • NewsAPI • MapboxGL • Google Cloud • AWS (basic)
LinkedIn: www.linkedin.com/in/ayub-khan-241b72354
Email: ayub2005.ak@gmail.com
