Tomas Pop

Machine Learning Engineer | MSc. Graduate (Charles University)

View Proprietary Industry Projects (Cisco)

LLM Pipeline for Rapid Email Phishing Detector Development

Built a modular library for automated labeling and rapid text-based classifier (detector) deployment. The detector developed via this procedure processed 550k+ weekly detections at the time of development.

Transformers ONNX OpenAI AWS
High-Throughput ML Detectors Development and Inference Optimization

Developed (some of the) ML-based detectors for the email anti-phishing engine.

Reduced existing ML-based detectors' inference latency for the proprietary email anti-phishing engine, protecting hundreds of millions of emails weekly via PyTorch, ONNX, and CUDA.

PyTorch ONNX CUDA GPU Acceleration

"Music of the past" (Archived Projects and Experiments)

Text-based Retrieval on Enron Email Dataset

Information retrieval - visualizing emails (high-dimensional semantic embeddings) via dimensionality reduction (UMAP, k-means).

Python spaCy Bokeh Dimensionality Reduction
ToTheMoon - Cryptocurrency Trading Bot

High-performance cryptocurrency trading bot (C++) with C# bindings focusing on latency and technical indicators.

C++ C# FinTech
Discord Crypto Alert Bot

Real-time cryptocurrency price alerts for Discord.

Java