Large Language Models (MLLM/LLM): 1. Autolabeling Pipeline: Designed and implemented an automated labeling pipeline using MLLM to streamline data preprocessing. 2. Retrieval-Augmented Generation (RAG) System: Built a comprehensive RAG system integrating voice, document, and text inputs with voice and text outputs utilizing MLLM, LLM, RAG, TTS (Text-to-Speech), and STT (Speech-to-Text) technologies. Computer Vision (CV): 1. Photo Validation Pipeline: Engineered a pipeline for validating photo quality and accuracy. 2. Photo Search Engine: Created an advanced search engine for efficient image retrieval based on various criteria. Performance Optimization: 1. VLLM Integration: Implemented VLLM to accelerate local LLM and MLLM operations, enhancing processing speeds. Backend Integration and API Development: 1. FAPI Utilization: Developed RESTful APIs using FAPI for seamless backend integration and communication between front-end applications and machine learning models. Data Engineering and Validation: 1. LangChain for Data Mining and Validation: Utilized LangChain to build intelligent agents for data mining and validation tasks, ensuring data integrity and efficient processing workflows. General Skills: 1. Integrated models using Docker, FastAPI, and Aiogram. 2. Optimized system architecture for scalability. 3. Translated business requirements into mathematical models and solutions. Technologies Used: - Languages & Frameworks: Python - Machine Learning: MLLM, LLM, RAG, VLLM, LangChain, Faiss, vllm - Computer Vision: OpenCV, PyTorch, Image, ONNX - APIs: RESTful APIs, FAPI - Other Tools: TTS/STT technologies, Whisper/WhisperX

Computer Vision: 1. Developed a comprehensive 4-stage face recognition system from scratch 2. Optimized and ported model inference from Python to Rust 3. Designed and implemented model architecture (custom or fine-tuned) 4. Created efficient data annotation systems 5. Implemented visual estimation of goods 6. Performed model conversion (e.g., to ONNX format) 7. Developed and validated datasets, including data mining techniques Large Language Models: 1. Constructed LLM systems incorporating RAG, Agents, Voice-to-Text, and Text-to-Voice 2. Implemented and optimized local LLM systems deployments 3. Evaluated RAG systems with RAGAs/custom metrics Keywords: Computer Vision / Python (Programming Language) / ML System Design / FastAPI / Docker / PyTorch / Machine Learning / Research and Development (R&D) / Large Language Models (LLM) / Natural Language Processing (NLP)

- Developed computer vision models to solve various tasks and improve visual recognition capabilities. - Implemented OCR projects to extract text and data from documents, enabling efficient document processing. - Conducted business analysis and utilized machine learning techniques to optimize operations and improve business performance. - Created APIs using FastAPI framework to provide efficient and scalable interfaces for data and functionality access. - Managed databases, particularly PostgreSQL, ensuring proper storage, retrieval, and organization of data. - Developed Telegram Bots using the aiogram framework, enabling seamless interaction and automation on the Telegram platform. keywords: Computer Vision Models / OCR Projects / Document Processing / Business Analysis / Machine Learning / API Development / FastAPI / Databases / PostgreSQL /Telegram Bots / aiogram / Classical Machine Learning / Scrapers / Regex

- Developed and optimized computer vision models for various tasks including action recognition, face detection/recognition, object detection/tracking, and OCR - Implemented data pipelines (ETL) and developed database architectures using PostgreSQL - Created backend systems using Go and Python, incorporating JWT authentication and database integration - Developed APIs using FastAPI and Gin frameworks for efficient data and functionality access - Collaborated with frontend/backend developers and UX/UI designers on project development - Conducted system and business analysis using tools like UML, Miro, and DBeaver - Developed frontend applications using Flutter (Dart) for web and Android platforms - Worked with video codecs for encoding and decoding, and developed Video Control Systems - Converted models to ONNX format and deployed them on TPU - Implemented Telegram Bots using the aiogram framework - Managed databases including PostgreSQL, MySQL, and SQLite - Utilized machine learning techniques to optimize business operations and improve performance - Developed scrapers and implemented classical machine learning algorithms

- Developed projects for optimizing production processes. - Conducted technological calculations, analytics, and forecasting. - Performed substance balance calculations. - Analyzed the condition of dynamic equipment based on its parameters. - Ensured compliance with safety and fire safety regulations. - Integrated with other production departments/units. - Worked with process flow diagrams. - Operated and maintained process equipment. Keywords: Production Process Optimization, Technological Calculations, Analytics, Forecasting, Substance Balance, Dynamic Equipment Analysis, Safety Compliance, Fire Safety, Process Flow Diagrams, Process Equipment. Skills: Python / MySQL / AutoCAD / Excel / Analysis / Safety Compliance / Process Flow Diagrams / Equipment Operation

Нефтихимия, ТНГ

Stepik