ALEKSANDR NAGAEV
] LinkedIn | a GitHub | û Telegram | ç Website
PROFILE
ML R&D Team Lead with 5+ years of expertise in NLP and Computer Vision, dedicated to advancing state-of-the-art AI
through novel algorithms and architectures. Proven ability to lead research projects from conception to publication and deliver
innovative solutions for complex applied problems. Strong background in both theoretical research and practical, optimized
deployment.
EXPERIENCE
Team Lead, RnD Vision June 2020 – Present
SberAI Moscow, Russia
• Led the research and development of cutting-edge multimodal and computer vision models, delivering SOTA results.
• Multimodal LLM Project: Improved model performance by 5-10% on vision-centric benchmarks by integrating novel
modalities and cross-modal alignment techniques.
• RSL Project: Developed the world’s rst general-domain Russian Sign Language (RSL)-to-text translation model, enabling
real-time interpretation for broad accessibility applications.
• SLOVO Framework: Led the creation of the world’s rst open-source framework for isolated sign language recognition.
Achieved SOTA results: 65.3% accuracy on WLASL (American SL) and 87.3% on the proprietary RSL benchmark. Research
published at ICVS’23 and presented at a CVPR’23 workshop.
• Real-time Talking Head Assistant: Architected an end-to-end system for a video conferencing chatbot, integrating a
spotter, ASR-API, RuGPT-3, a proprietary text2landmark model for lip-sync, and a head animation system.
• Dynamic Gesture Recognition: Designed an algorithm for device control using heuristics and hand detection models,
enabling precise gesture-based interaction.
• Demo Stand for Sign Language Teacher: Built a real-time system recognizing 1000+ gestures on CPU, showcased at
international conferences (ICTWEEK’24, GITEX’24).
• Facial Enhancement System: Developed a real-time 3D morphing and skin smoothing pipeline achieving 30 FPS on
mid-tier CPUs using SSD face detection and OpenCV optimizations.
Data Analyst October 2019 – June 2020
Active Business Consult Moscow, Russia
• Designed an error correction system for dialogue interfaces using synthetically generated ASR errors, achieving 15% WER
reduction via a context-aware ELMo model.
TECHNICAL SKILLS
Languages & Frameworks Python, PyTorch, OpenCV, Pandas, Hugging Face
ML Domains NLP: Transformers, LLMs, MultiModal LLMs
Computer Vision: Image/Video Processing, Sign Language Recognition, Gesture Recognition
Deployment & Tools Docker, Git, CI/CD, ONNX, CPU/Edge Optimization
Languages Russian (Native), English (Intermediate - B1)
PUBLICATIONS AND PRESENTATIONS
Research Proles
• Google Scholar
• Orcid
• Scopus
Articles
• Habr
Technical Talks
• Saint HighLoad++ 2024 l
• GIGA RnD Day 2024 l
• HighLoad++ 2023 l
• X5 Data science meetup 2023 l
EDUCATION
Master of Data Science, Ural Federal University 2019 – 2022
Instructed undergraduate students in Machine Learning and Python programming.
Bachelor of Computer Science, Ural Federal University 2015 – 2019
Led cross-functional teams in semester-long academic projects using Agile methodology.
