Dmitriy Konyrev
Verified Expert in Engineering
ML Engineer and Developer
Barcelona, Spain
Toptal member since July 13, 2026
Dmitriy is an AI engineering leader with 12 years of software experience and 9 years of experience leading ML teams. He builds production LLM applications, AI agents, evaluation systems, RAG pipelines, and scalable ML infrastructure. At SuperAnnotate, his team reduced manual annotation effort by over 40% and accelerated QA by up to 100x. Dmitriy now serves as CTO of Argmin AI, helping companies turn AI prototypes into reliable production systems.
Portfolio
Experience
- Machine Learning - 10 years
- Natural Language Processing (NLP) - 10 years
- PyTorch - 9 years
- Team Management - 9 years
- Machine Learning Operations (MLOps) - 7 years
- LLM Evaluation - 3 years
- Large Language Models (LLMs) - 3 years
- AI Agents - 2 years
Preferred Environment
Kubernetes, MLflow, Jenkins, Docker, TeamCity, Agentic AI, AWS IoT
The most amazing...
...AI system I've built automated legal document analysis and helped cut the time to make corporate loan decisions from 7 days to 7 minutes.
Work Experience
CTO
Argmin AI
- Built and shipped Judge Builder, a guided tool for creating and calibrating LLM-as-a-judge evaluators from examples and expert feedback.
- Developed the Argmin AI platform to evaluate LLM applications, compare quality and cost across versions, and support cost-optimization decisions.
- Designed Argmin AI's optimization engine to reduce LLM inference costs by 2-5x through smarter model choice, prompt strategy, and routing.
ML Team Manager
SuperAnnotate
- Managed an international, cross-functional team of 3 ML engineers and one product manager, focused on productized delivery of annotation and evaluation tools.
- Launched automatic annotation tools for images, videos, and text, helping regain feature parity with competitors and significantly reduce manual labeling effort and turnaround time.
- Built semantic data curation features using embeddings (BLIP, OpenCLIP, and PyTorch metric learning) and OpenSearch vector DB, enabling users to manage datasets more effectively and manually select similar objects for labeling or review.
- Developed LLM evaluation pipelines using DeepEval and DeepTeam, including LLM-as-a-judge scoring with integrated guardrails, automating QA workflows, and improving customer confidence in label quality.
- Created agentic pipelines for pre-annotation using automatic and semi-automatic feedback loops, improving annotation accuracy and reducing human-in-the-loop load.
- Established an MLOps pipeline in collaboration with SRE and QA teams using AWS, MLflow, and Jenkins, automating quality checks and reducing deployment overhead.
- Led internal research on detecting AI-generated text and identifying impactful fine-tuning samples for non-STEM datasets, resulting in a benchmarked model published on Hugging Face using TRL, PEFT, BitsAndBytes, Ollama, and DeepEval.
Risk Modeling Team Lead | Manager of ML Engineering
Bank Dom.RF
- Managed and monitored a team of 5 risk-oriented ML engineers.
- Introduced bureau-based credit scoring models for individuals, entrepreneurs, and SMEs, enabling data-driven loan approval and improved risk assessment across multiple client segments.
- Designed and launched a real estate scoring service for B2B clients, automating creditworthiness evaluation for construction and property companies and reducing manual review time.
- Delivered a cash-flow forecasting model for mortgage securitization, improving portfolio transparency and supporting risk-based pricing decisions.
R&D Team Lead
Sberbank
- Managed a hybrid product plus research group of 6 data scientists and 5 engineers, adapting bleeding-edge NLP/CV (Transformers and BiLSTM) to enterprise use cases.
- Built an NER service for legal documents used in the online loan pipeline, enabling Sberbank's "Seven Minutes" decisioning for large corporate loans that previously took 7 days.
- Designed and shipped Digital Manager, an email and document assistant that recommended correct actions in 90% of routine executive cases.
- Built delegation recommendations for the Digital Manager, suggesting effective delegation schemes in 75% of executive workflow cases.
ML Engineer
Mail.ru Group (myTarget)
- Improved CTR prediction by incorporating user clickstream and session features into ranking models, enhancing ad relevance for banner placements.
- Supported deployment of a C++ CTR prediction service for high-load ad ranking, helping move model improvements into production banner placement workflows.
- Analyzed how search-session context, banner context, and user click history influenced click relevance, informing feature design for CTR ranking models.
Software Engineer
KB Panorama
- Contributed to a navigation system for cargo vessels, implementing features and fixes across routing, UI, and data processing modules.
- Optimized vessel-routing algorithms reduce the time required to calculate routes for cargo-ship navigation workflows.
- Fixed bugs across routing logic, user interface behavior, and data-processing components, improving the reliability of the navigation system.
Experience
Automatic Judge Builder
https://argminai.com/ai-quality-without-ml-teamEmbeddingStudio
https://github.com/EulerSearch/embedding_studioGenerated Text Detector
https://huggingface.co/SuperAnnotate/ai-detectorImage Auto Annotation Tools
https://www.superannotate.com/image-annotation-toolVideo Auto Annotation Tools
https://www.superannotate.com/video-annotationEducation
Master's Degree in Applied Mathematics and Computer Science
Moscow Power Engineering Institute - Moscow, Russia
Bachelor's Degree in Applied Mathematics and Computer Science
Moscow Power Engineering Institute - Moscow, Russia
Skills
Libraries/APIs
PyTorch, NumPy, Pandas, Scikit-learn, SciPy, Matplotlib, Gradio, OpenCV, PaddleOCR, Vue, React, Hugging Face Transformers, OpenAI API, Node.js
Tools
Amazon OpenSearch, Jenkins, TeamCity, TensorBoard, Plotly, Seaborn, Celery, You Only Look Once (YOLO), Torchvision, Amazon SageMaker, Docker Compose, Segment Anything Model (SAM), Terraform
Languages
Python, Clipper, JavaScript, TypeScript, C++, Python 3
Frameworks
Agentic Frameworks, Alembic, LlamaIndex, Flask, Boost, Qt, Next.js, Hadoop, LangGraph
Industry Expertise
Project Management
Storage
Data Pipelines, Amazon S3 (AWS S3), PostgreSQL, MongoDB, Amazon DynamoDB, Elasticsearch, Redis
Platforms
Docker, Google Cloud Platform (GCP), AWS IoT, Ollama, Kubernetes, Langfuse, NVIDIA CUDA, Amazon EC2, Amazon Web Services (AWS)
Paradigms
Model Context Protocol (MCP)
Other
Natural Language Processing (NLP), Machine Learning, AI Agents, Agentic AI, Artificial Intelligence (AI), Retrieval-augmented Generation (RAG), Data Science, IT Strategy, Multi-agent Orchestration, Model Monitoring, Model Deployment, Model Evaluation, Model Tuning, Agentic AI Systems, LLM Integration, Debugging, Orchestration, Architecture, Leadership, Multi-agent Systems, Software Architecture, Team Leadership, Technical Leadership, Product Management, System Design, Agentic Workflow Design, RAG Pipelines, RAG Systems, Agentic RAG Systems, RAG Architecture, Large Language Models (LLMs), LLM Evaluation, Metric Learning, Risk Modeling, Information Extraction, Computer Vision, LangChain, FastAPI, EC2, Fine-tuning, Data Scientist, Credit Scores, Large Language Model Operations (LLMOps), Statistical Modeling, BLIP, DeepEval, MLflow, PEFT, Transformers, Sentence Transformers, FAISS, Pgvector, Qdrant, ChromaDB, Ragas, Detectron2, ECR, NVIDIA NIM, ONNX Runtime, TimescaleDB, Mathematical Statistics, Hugging Face, Machine Learning Operations (MLOps), NVIDIA Triton, Vector Databases, DeepTeam, RAGAS, BitsAndBytes, Financial Modeling, Team Management, Data Collection, Data Annotation, Milvus, Libraries, Triton Inference Server, YOLOv5, LoRa, Amazon Bedrock AgentCore, CI/CD Pipelines, Quantization, Recommendation Systems, Startups, CTO
How to Work with Toptal
Toptal matches you directly with global industry experts from our network in hours—not weeks or months.
Share your needs
Choose your talent
Start your risk-free talent trial
Top talent is in high demand.
Start hiring