Open to opportunities

Sergey Orlov

@sergeyorlov

Message

Data Scientist at Cashco Financial who built a credit decisioning platform saving approximately $200K annually.

Canada

Message

At Cashco Financial, I architected and built an end-to-end credit decisioning platform, replacing a third-party engine with a containerized FastAPI service on Azure. It processes approximately 14,000 underwriting decisions monthly and saves approximately $200K annually.

I also built cash-flow underwriting that reduced credit-bureau pulls sevenfold while maintaining decision quality for thin-file and first-time borrowers. Earlier, I led data science and HPC teams at National Research Tomsk State University and developed production AI services, including a voice assistant and a recommendation system.

Experience

Work history, roles, and key accomplishments

Current

Data Scientist

Current

Cashco Financial

Apr 2024 - Present (2 years 5 months)

Architected and built an end-to-end credit decisioning platform, replacing a third-party engine with a containerized FastAPI service on Azure. Designed and productionized five XGBoost probability-of-default models and rebuilt credit approval policy, improving decision quality and reducing defaults.

XGBoost fastAPI Azure Python SQL SHAP Airflow DBT Metabase OCR LLM MCP Credit Risk Modeling Underwriting Model Governance CI CD Docker Git

Data Scientist

Cybera Inc

Oct 2023 - Mar 2024 (5 months)

Designed and deployed an entity-resolution pipeline for large unstructured datasets, achieving 92% classification accuracy with BERT/LLaMA2 embeddings and Siamese neural networks. Scaled pipeline execution on GCP using Docker, multiprocessing, GPU acceleration, and a 24-node virtual-machine cluster.

BERT LLaMA 2 Siamese Neural Networks scikit learn TensorFlow RAG Faiss GPT 4 Docker GCP Multiprocessing GPU Virtual Machines

Director of Data Science & HPC

National Research Tomsk State University

May 2016 - Dec 2022 (6 years 7 months)

Established and led a Data Science research laboratory, translating applied ML research into production services and mentoring 30+ students. Managed 12 engineers supporting a Platform-as-a-Service environment for 5,000+ HPC users and led production AI services including a voice assistant and a recommendation system.

Data Science Machine Learning HPC Platform as a Service Team Leadership Mentorship Voice Assistants Recommendation Systems

Computational Modelling & ML Researcher

National Research Tomsk State University

Feb 2012 - May 2016 (4 years 3 months)

Developed predictive-modeling software in C++ and parallel computing (OpenMP, MPI) for large-scale scientific simulations. Applied neural networks and genetic algorithms to optimization problems including aerodynamic UAV design, and optimized Linux/HPC workflows on a 500+ node cluster.