Pavel Shibanov

• Managed projects end-to-end, from initiation through closure. • Designed, planned, deployed production LLMs on GKE Autopilot and AWS SageMaker. • Trained, and integrated 3 PyTorch LSTM binary classifiers into an existing FastAPI and React application. • Fine-tuned Falcon-40B to financial data sources for sentiment analysis with LoRa. • Performed analysis and visualization using Pandas, Matplotlib, and TF Data Validation. Tech: Python, TensorFlow, PyTorch, Vertex AI, Typescript, SQL, NoSQL, Kubernetes.

Responsibilities: For a global fashion trading platform implemented backend functionality with Express.js and Sequelize, implemented frontend functionality with React, developed ETL solution that transformed excel documents with ExcelJS, AWS Lambda function and AWS S3, developed CI pipeline with Github Actions, updated documentation. Technologies: Node.js, Express.js, PostgreSQL, Sequelize ORM, ExcelJS, React, ECS, AWS Lambda, Github Actions

Implementing, deploying and maintaining API for exchanging information with over 3000 banks and financial institutions, normalizing access and removing inconsistencies in communication. Technologies: K8s, helm, Ansible, Docker, AWS, Express.js, GitLab, Node.js, JavaScript, TypeScript, puppeteer.

Implementing new features, refactoring, reviews, lib updates. Tracking user interaction become my obsession. I always think how to measure something useful, how to analyze something new. Trying to write effective and compact code, that anyone can understand and maintain.

I was learning enterprise tooling and scientific approach to software development. My day-to-day job was implementing new functionality using AngularJS, AngularUI Bootstrap, Angular UI Router Leaflet.js, D3.js and a lot of other libraries. My code was reviewed by professionals with awesome level of expertise in front-end and back-end. I learned how important to write tests, how advanced tooling and unique developer environment saves great amount of time.

Web-development