Professional SummaryResults-driven Python developer and Machine Learning engineer with hands-on experience building real-world AI systems. Specialized in NLP, text classification, and continual learning pipelines. Proven track record of delivering production-ready ML solutions — from data collection and model training to deployment in desktop applications used by emergency services.Participated in AI/ML seminars and actively follow the latest research in the field.Work ExperienceSoftware Engineer & ML DeveloperMinistry of Emergency Situations (MCHS) — Emergency Response System2023 – 2024Designed and developed a full-stack desktop application for emergency response teams:
Built an NLP classification pipeline using ruBERT (fine-tuned) to automatically categorize Telegram messages into emergency categories: earthquakes, floods, fires, accidents, etc.Implemented real-time data collection from Telegram channels using Pyrogram (Telegram API) with automatic text cleaning (removal of noise, special characters, emojis) to ensure clean training dataDeveloped a continual learning system — the model retrained automatically on new data volumes without full retraining from scratchAdded on-demand browser-based training trigger — operators could initiate model retraining directly from the UIIntegrated classification results into an interactive map (geo-visualization) for field rescue teamsBuilt the desktop UI Exported models using ONNX Runtime for optimized inference
Stack: PyTorch · HuggingFace Transformers · BERT · Natasha (NER) · scikit-learn · HuggingFace Datasets · Pyrogram · Pandas · NumPy · Kivy · KivyMD · ONNX RuntimeSoftware
Developed and maintained internal software tools for the networkBuilt data processing scripts and automation pipelinesApplied basic ML techniques for sales pattern analysis and demand forecasting using scikit-learn and Pandas
SkillsMachine Learning & AIPyTorch · HuggingFace Transformers · scikit-learn · ONNX Runtime · HuggingFace Datasets · Continual Learning · Model Fine-tuningNLPruBERT · Text Classification · Named Entity Recognition (NER) · Natasha · Tokenization · Data CleaningDataPandas · NumPy · Data PipelinesParsing & APIsPyrogram (Telegram API) · newspaper3k · feedparser · BeautifulSoup · requestsDesktop DevelopmentKivy · KivyMD · Python desktop appsOtherGit · REST APIs · JSON · SQLiteEducation & Development
Participant — AI/ML industry seminarsSelf-directed learning: NLP, transformer architectures, production ML systems
What I Can Do For You on Guru
Train or fine-tune NLP models (text classification, NER, sentiment analysis)Build Telegram bots and data scrapersCreate desktop applications with ML integrationClean and prepare datasets for trainingBuild automation scripts in PythonIntegrate ML models into existing apps (ONNX, REST API)
Languages
Russian — NativeEnglish — just to review the documentation ,Italian — Conversational