Stephan Leznikov

About Me

Hello there! I'm Stephan, a Computer Science and AI student at Queen's University with a hands-on approach to building intelligent systems. Currently, I'm working as an AI/Machine Learning Intern at Kyndryl, where I designed and developed an AI-powered document validation and data extraction system that automated a client's mortgage application review process, effectively doubling their application throughput. From deep learning models and NLP pipelines to full-stack web apps, I love taking an idea from concept to production. I'm always looking for the next challenging problem to solve, whether that's training a neural network, architecting a clean API, or collaborating with a team to ship something meaningful.

Experience

Skills

Education

Kyndryl - AI / Machine Learning Intern
May 2025 – August 2026

Designed and developed an AI-powered document validation and data extraction system for a client's mortgage application process. The solution automated the review of supporting documents submitted by applicants, significantly reducing manual workload for reviewers and enabling the client to effectively double their mortgage application throughput.

Engaged directly with the client throughout the project lifecycle, gathering requirements, presenting progress updates, and iterating on deliverables to ensure alignment with business objectives.

Programming Languages
Strong command of Java, Python, JavaScript/Node.js, C++, and R with a focus on clean, object-oriented architecture and scalable design patterns.
AI & Machine Learning
Hands-on experience building and fine-tuning deep learning models, leveraging frameworks such as TensorFlow, PyTorch, and Hugging Face Transformers. Proficient in LLM integration, prompt engineering, and deploying AI-driven solutions in production environments.
Full-Stack Web Development
End-to-end development of responsive web applications using HTML, CSS, JavaScript, and modern frameworks. Strong eye for intuitive UI/UX design and performance optimization.
Data & Cloud Infrastructure
Proficient in SQL (MySQL, PostgreSQL) for complex data modeling and querying. Experienced with cloud platforms and CI/CD workflows for reliable, automated deployments.
Developer Tools & Workflow
Skilled with Git version control, RESTful API design, Agile methodologies, and collaborative development in cross-functional teams.

Queen's University
Bachelors of Science
Computer Science / Artificial Intelligence
2022 - Current

Portfolio

Document Validator & Extractor

AI-powered document validation and data extraction system deployed on Azure Functions. Processes documents and transmits extracted data to Salesforce for centralized management.

Queen's AutoDrive

Played a crucial role in detecting and addressing potential system vulnerabilities by simulating complex real-world situations, ensuring the machine learning model's reliability in a variety of conditions.

Food Classifier

Deep learning model using MobileNetV2 transfer learning to classify 41 food categories with 85-90% accuracy. Features data augmentation, GPU optimization, and TensorBoard monitoring.

StatZone

Sports analyst chatbot utilizing the OpenAI API for real-time insights into team performances, player statistics, and game strategies. Integrated advanced machine learning models and data analytics to predict outcomes.

Sentiment Analysis

Advanced NLP system combining Random Forest ML with rule-based analysis. Features negation detection, intensity modifiers, and sentiment lexicon integration for nuanced text classification.

2D Platformer

A 2D platformer game built with Unity and C#, featuring custom shaders and platform mechanics.