Polina Stina

AI & Machine Learning student with hands-on experience building ML and LLM-based systems, including document classification, semantic retrieval, embeddings, and prompt-driven pipelines. Experienced with Python, PyTorch, HuggingFace, and end-to-end ML workflows from preprocessing to inference and evaluation. Also built and deployed a full-stack automated data pipeline (Python, FastAPI, SQLite) for real-time business monitoring. Interested in applied AI, backend-integrated ML systems, and production-oriented development.

IELTS, Cambridge English B2 First (FCE), Machine Learning Specialization Andrew Ng, Coursera, Deep Learning Specialization Andrew Ng, Coursera, AI agents Andrew Ng, DeepLearning.AI

Jul 2026 - Aug 2026, Automated Chemical Price Tracking & Dashboard System, Python, SQLite, FastAPI, Playwright

• Built and deployed an automated price monitoring system for a chemicals business, scraping data from sci99.com using Playwright with network interception to capture data not available through a public API.
• Designed the full pipeline from automated data collection through to a live, queryable dashboard, replacing manual price checking with an always-up-to-date view.
• Live dashboard: chem-price-tracker.vercel.app
• Code: github.com/polsti/chem-price-tracker

May 2026 - June 2026, Job Offer Prediction System, Python, Scikit-learn, FastAPI

• Developed an ML system predicting job offer probability based on candidate profile features and technical stack.
• Implemented preprocessing, feature engineering, model training, and evaluation workflows.
• Built prediction API using FastAPI with structured JSON input/output.
• Applied classification models and feature importance analysis to interpret predictions.

Apr 2026 - May 2026, Robot Arm Control via Reinforcement Learning, Python, Gymnasium, MuJoCo, Reinforcement Learning

• Developed a reinforcement learning environment for robotic arm control using Gymnasium Robotics and MuJoCo.
• Worked with observation spaces, action spaces, reward functions, and goal-based RL environments.
• Implemented robotic manipulation workflows and reinforcement learning experimentation pipelines.
• Explored environment metrics, simulation-based training, and agent evaluation workflows.

Aug 2025 - Nov 2025, AI Document Trainer (LLM Pipeline), Python, HuggingFace, Qwen, Embeddings, NLP

• Developed an LLM-based document processing pipeline for structured summarization and conversational retrieval.
• Implemented chunking, embeddings, semantic search, and multi-step prompt workflows.
• Designed an end-to-end pipeline from preprocessing and embedding generation to chatbot-style interaction.
• Improved retrieval consistency through optimized chunking and retrieval settings.

Aug 2025 - Sep 2025, Document Image Classification System, Python, PyTorch, CNNs, Computer Vision

• Built a two-stage document image classification system using CNN-based architectures.
• Performed dataset preparation, augmentation, model training, evaluation, and architecture experimentation.
• Improved classification accuracy through augmentation strategies and iterative tuning.
• Worked with preprocessing pipelines and image-based inference workflows.

Apr 2024 - May 2024, TV Streaming Database Website, Flask, SQL, Python

• Developed a full-stack database website with Flask backend and SQL database integration.
• Implemented ER models, query logic, filtering, and backend data workflows.
• Built backend functionality for storing, retrieving, and displaying structured content.