Zhuikov Alexander
Viktorovich
Moscow
Highly experienced AI and Neural Network specialist with over 24 years of experience, specializing in the development and implementation of advanced AI models, including large language models. Proven ability to lead teams, manage complex projects, and deliver innovative solutions across various industries. Seeking to leverage extensive expertise in AI research and development to contribute to cutting-edge projects.
Scroll to explore
Experience
My professional journey and the amazing teams I've been part of
Nov 2018 — Invalid Date
Robots on the Moscow Exchange (Quik) and on crypto exchanges.
Experienced market maker.
IciviLab
Aug 2015 — Invalid Date
My main difference: I check all my work in practice, my GPUs are never idle and always train the model 24/7
Strong Artificial Intelligence of our own development (potential level higher than Llama2, Mixtral-8x7B and Mistral with the possibility of reaching the OpenAI level)
70Gb and there are many tiny versions of 8Gb, tailored to the task, system, production, strategy, Man
Models are analogues of ChatGPT in equipment and systems.
Super-AI development for more than 10 years.
State-of-the-art Development of an analogue of Chat GPT + implementer, based on the latest developments
Analyzed all the advantages, disadvantages of working with personnel from the hiring departments of large companies and the problems of higher educational institutions in countries (not only universities in the Russian Federation, but also the EU and the USA are becoming deeply outdated).
Know all the intricacies of how to gather hackers and directors into teams, where to find them, how to organize them (we will replace outdated interviews, incompetent and not exactly relevant tests, 'funnels', etc. outdated tools of departments, HR, company and government recruiting systems. structures.)
Vector database, Your personal search by _Meaning_, and not just by words
Optimized and compressed AI models, _for putting AI into small machines_
Quantize full Hugging Face models (LLama2 family) in GPTQ, GGUF so that they fit into GPUs with 16Gb - 100Gb.
Make neural networks, installations: in microcomputers, boards, drones, cars, cars, machine tools.
Develop automatically Superintelligent AI
Distribution of application and neural network flows to CPU and GPU accelerators with Geforce RTX and Tesla.
Model 11 times smaller than the one in the manual (count_params 148,645 versus 1,625,611) Superior in metrics.
Servers for trading robots and websites
Apr 1999 — Jul 2012
Icivi company, exchange investment department
Mar 1999 — May 2009
Education
Where I built the foundation of my expertise
Class of