Nikita Goryachev

Lead a dedicated division focused on advanced AI and ML solutions:

• Transformers implementation for RecSys.
• Large Language Models (LLMs) & RAG: Develop and deploy LLMs and Retrieval-
• Augmented Generation models for internal company applications.
• Speech Processing: Design and implement models for voice assistant support
• and voice transcription.
• Computer Vision (CV): Create computer vision models that generate video
• embeddings for the NUUM platform, enhancing recommendation systems.

Operational Responsibilities:

• Integrate developed models into scalable services.
• Manage GPU clusters to ensure efficient processing and resource utilization.
• Oversee MLOps and DevOps aspects of the machine learning infrastructure.
• Conduct initial research and experimentation to validate new ideas and
• approaches.
• Package and deploy models as reliable services.
• Performance Engineering:
• Estimate and analyze load requirements.
• Optimize models and services for GPU performance.

Testing & Quality Assurance:

• Execute integration and load testing to ensure system stability.
• Implement A/B testing strategies to validate model performance in production.
• Transition ML models from experimental phases to fully production-ready
• systems.

- Transformers optimization in recommender systems.
- Working with NLP/RecSys SOTA algorithms.
- Expertise in developing and managing NVIDIA A100 cluster infrastructures, contributing to the advancement of large-scale AI model training.
- Development of a comprehensive Software Development Kit (SDK) for RecSys.
- Scrum-Master: Managing and distributing the team's workload effectively.

- Software development for Neuro-Interface devices. Development of anomaly cleaning alghorithms and ML models for tremendous number of scenarios. Working with EEG data (electroencephalogram)

- ML models development for biggest retail of Russia and Kazachstan

- Work as a project manager in addition to my main role (Data Scientist)

Work projects:

• Python API dev for EEG device + ML implementation
• ML optimization for retail (One of the largest retailers in Russia)
• Anomaly Detection module in EEG data
• Fiscal (OFD) Big Data Pipelines develompment for BI's (Russia)
• NLP pipelines for Big Data (one of Kazachstan's biggest bank and Fiscal operator)

- Implementation of various ML methods, visualization.

- Weekly reports for the corporate business department.

- Worked at COVID headquarters. Sber launched a robot to collect information on the health status of employees. I did the analysis to identify the "green", "yellow" and "red" zones to define triggers of disease and to try to prevent early severe disease and potential death.- Implementation of various ML methods, visualization. - Weekly reports for the corporate business department. - Worked at COVID headquarters. Sber launched a robot to collect information on the health status of employees. I did the analysis to identify the "green", "yellow" and "red" zones to define triggers of disease and to try to prevent early severe disease and potential death.