Research Scientist at JetBrains and former Head of Applied ML at Moonvalley.ai
Montenegro
Ivan
Drokin
Council Member
Ivan Drokin is a research scientist at JetBrains with over a decade of experience in artificial intelligence, computer vision, and deep learning, currently based in Budva, Montenegro. Throughout his career, he has built and led high-performing teams, established successful AI-driven companies, and contributed to breakthrough technologies in generative AI, medical imaging, and video generation.
Since March 2025, Ivan Drokin has served as a Research Scientist at JetBrains, the Czech software development company that serves over 15.9 million developers worldwide and generates revenue exceeding $400 million annually. In this role, he defines the research agenda around Diffusion World Models, focusing on model design, training procedure development, and data requirements gathering. His primary research centers on developing world model diffusion models capable of processing multimodal control signals, including text and agent actions.
From July 2024 to September 2025, Ivan Drokin held the position of Head of Applied ML at Moonvalley.ai, an artificial intelligence research company specializing in generative video tools that has raised over $110 million in investments. He led the applied machine learning team, managing planning, team development, hiring, and employee growth while handling communication with C-level executives and stakeholders to define team objectives. His team developed and maintained the machine learning components of data processing pipelines, including quality assessment, cleaning, deduplication, and caption enrichment. He worked closely with the product team and partners to expand the capabilities of the Marey fundamental model, which generates high-quality videos up to 30 seconds long with precise detail control.
From January 2023 to June 2024, Ivan Drokin served as Head of Data Science at Prequel Inc, an American technology company developing mobile applications for photo and video editing using artificial intelligence. He led a cross-functional team of more than 11 deep learning researchers, MLOps engineers, and data acquisition engineers to research, implement, maintain, and develop high-load image and video processing, stylization, and generation backend services. Under his leadership, he stayed current with the latest advancements in generative AI, provided visionary direction for the research and development team, and controlled projects from inception to completion. He conducted cutting-edge research in generative AI, explored novel algorithms for image, video, and text stylization and generation, and implemented prototypes to validate the feasibility of new models for emerging products and use cases. He trained and fine-tuned deep learning models, evaluated emerging technologies, and presented new possibilities to C-level executives, the Board of Directors, and the Product Team while ensuring compliance with relevant regulations and standards.
In August 2016, Ivan Drokin co-founded BrainGarden.ai, serving as Chief Science Officer and General Director until December 2022. The company provided research and development as a service, focusing on state-of-the-art computer vision and AI solutions for clients including Arrival, LG, and Nyenschanz Automatica. He built the company into a profitable enterprise that reached revenue of 45 million rubles in 2021. As co-founder, he contributed expertise in deep learning, machine learning, computer vision, and applied mathematics while building and managing a team of deep learning researchers. He defined company strategy, participated in sales as a technical expert, communicated with clients to ascertain project goals and gather requirements, designed research projects including model selection and evaluation processes, developed computer vision models for industrial and consumer electronics applications, and engaged in public activities to promote the company brand.
Ivan Drokin also co-founded Botkin.ai in July 2017, serving as Chief Research Officer until August 2022. The medical innovation company developed a platform for medical image analysis using artificial intelligence, raising over 250 million rubles in total investments and earning 51.1 million rubles in revenue by 2021. He joined the company to lead the Research and Development division with the objective of building an AI-powered platform for medical image analysis. He developed pipelines for instance and semantic segmentation, object detection, and classification in medical images while leading and mentoring machine learning and medical teams. He oversaw the entire machine learning model lifecycle, designed and implemented protocols for evaluating computer vision models at all stages including training, validation, and production monitoring, and communicated with research centers globally to establish partnerships and joint research initiatives. He established and reviewed data acquisition, gathering, and labeling protocols, including commercial partnerships with companies and clinics, built and managed annotation teams, prepared requirements for annotation protocols, and designed annotation tools to enhance process efficiency. He assisted Business and Product Teams in formulating new research areas, designed lean proof-of-concept projects for prospective products, and presented the company's products and technologies at professional conferences and meetups. Under his technical leadership, Botkin.ai secured partnerships with leading research institutes and companies including AstraZeneca, Lomonosov Moscow State University, Russian Research Center for Radiology, and Hospital Israelita Albert Einstein in São Paulo, Brazil, and acquired the CE Mark in 2021 for international expansion.
Since March 2025, Ivan Drokin has served as a Research Scientist at JetBrains, the Czech software development company that serves over 15.9 million developers worldwide and generates revenue exceeding $400 million annually. In this role, he defines the research agenda around Diffusion World Models, focusing on model design, training procedure development, and data requirements gathering. His primary research centers on developing world model diffusion models capable of processing multimodal control signals, including text and agent actions.
From July 2024 to September 2025, Ivan Drokin held the position of Head of Applied ML at Moonvalley.ai, an artificial intelligence research company specializing in generative video tools that has raised over $110 million in investments. He led the applied machine learning team, managing planning, team development, hiring, and employee growth while handling communication with C-level executives and stakeholders to define team objectives. His team developed and maintained the machine learning components of data processing pipelines, including quality assessment, cleaning, deduplication, and caption enrichment. He worked closely with the product team and partners to expand the capabilities of the Marey fundamental model, which generates high-quality videos up to 30 seconds long with precise detail control.
From January 2023 to June 2024, Ivan Drokin served as Head of Data Science at Prequel Inc, an American technology company developing mobile applications for photo and video editing using artificial intelligence. He led a cross-functional team of more than 11 deep learning researchers, MLOps engineers, and data acquisition engineers to research, implement, maintain, and develop high-load image and video processing, stylization, and generation backend services. Under his leadership, he stayed current with the latest advancements in generative AI, provided visionary direction for the research and development team, and controlled projects from inception to completion. He conducted cutting-edge research in generative AI, explored novel algorithms for image, video, and text stylization and generation, and implemented prototypes to validate the feasibility of new models for emerging products and use cases. He trained and fine-tuned deep learning models, evaluated emerging technologies, and presented new possibilities to C-level executives, the Board of Directors, and the Product Team while ensuring compliance with relevant regulations and standards.
In August 2016, Ivan Drokin co-founded BrainGarden.ai, serving as Chief Science Officer and General Director until December 2022. The company provided research and development as a service, focusing on state-of-the-art computer vision and AI solutions for clients including Arrival, LG, and Nyenschanz Automatica. He built the company into a profitable enterprise that reached revenue of 45 million rubles in 2021. As co-founder, he contributed expertise in deep learning, machine learning, computer vision, and applied mathematics while building and managing a team of deep learning researchers. He defined company strategy, participated in sales as a technical expert, communicated with clients to ascertain project goals and gather requirements, designed research projects including model selection and evaluation processes, developed computer vision models for industrial and consumer electronics applications, and engaged in public activities to promote the company brand.
Ivan Drokin also co-founded Botkin.ai in July 2017, serving as Chief Research Officer until August 2022. The medical innovation company developed a platform for medical image analysis using artificial intelligence, raising over 250 million rubles in total investments and earning 51.1 million rubles in revenue by 2021. He joined the company to lead the Research and Development division with the objective of building an AI-powered platform for medical image analysis. He developed pipelines for instance and semantic segmentation, object detection, and classification in medical images while leading and mentoring machine learning and medical teams. He oversaw the entire machine learning model lifecycle, designed and implemented protocols for evaluating computer vision models at all stages including training, validation, and production monitoring, and communicated with research centers globally to establish partnerships and joint research initiatives. He established and reviewed data acquisition, gathering, and labeling protocols, including commercial partnerships with companies and clinics, built and managed annotation teams, prepared requirements for annotation protocols, and designed annotation tools to enhance process efficiency. He assisted Business and Product Teams in formulating new research areas, designed lean proof-of-concept projects for prospective products, and presented the company's products and technologies at professional conferences and meetups. Under his technical leadership, Botkin.ai secured partnerships with leading research institutes and companies including AstraZeneca, Lomonosov Moscow State University, Russian Research Center for Radiology, and Hospital Israelita Albert Einstein in São Paulo, Brazil, and acquired the CE Mark in 2021 for international expansion.
From February 2015 to June 2016, Ivan Drokin worked at BIOCAD, a Russian biotechnology company with more than 2,700 employees engaged in pharmaceutical and biopharmaceutical research, development, and production. He applied machine learning algorithms, optimization methods, and bioinformatics to tasks related to pharmaceutical discovery and production process optimization. He developed mathematical models of cell lines to optimize productivity, designed and evaluated deep learning models for small molecule property prediction, and researched and implemented drug portfolio augmentation systems based on deep learning and genetic algorithms.
Ivan Drokin has received numerous professional recognitions throughout his career. He earned the Best Talk Award at AIST 2019 and the Best Paper Award at AIST 2020. Under his leadership, Botkin.ai received multiple awards including three honors at Digitalhealthawards 2019—"Breakthrough of the Year" in the Technology category, "Organization of the Year" in the Technology category, and the Grand Prix "Startup of the Year." The company also earned "Organization of the Year" in the Technology category (first place) and Research category (third place) at Digitalhealthawards 2020, received "The Technological Breakthrough Award" in 2021, won the AI Russia award in 2021, and placed in the top 10 at the Entrepreneurship World Cup in December 2021. Botkin.ai also achieved second place in the Evercare ranking of AI startups in Healthcare and won the AstraZeneca-Skolkovo Startup Challenge in 2019 and Patents Power 2019.
Ivan Drokin holds four patents related to medical AI and deep learning applications. These include a method for combining patient information through deep learning models into unified representations for diagnostics and prognosis, an invention covering patient data representation and transfer for machine learning algorithm analysis, a comprehensive patent covering the entire process from data acquisition to final machine learning predictions for diagnosing organ pathologies, and an approach based on point-cloud representation of human body CT scans with associated models and technologies.
Ivan Drokin has received numerous professional recognitions throughout his career. He earned the Best Talk Award at AIST 2019 and the Best Paper Award at AIST 2020. Under his leadership, Botkin.ai received multiple awards including three honors at Digitalhealthawards 2019—"Breakthrough of the Year" in the Technology category, "Organization of the Year" in the Technology category, and the Grand Prix "Startup of the Year." The company also earned "Organization of the Year" in the Technology category (first place) and Research category (third place) at Digitalhealthawards 2020, received "The Technological Breakthrough Award" in 2021, won the AI Russia award in 2021, and placed in the top 10 at the Entrepreneurship World Cup in December 2021. Botkin.ai also achieved second place in the Evercare ranking of AI startups in Healthcare and won the AstraZeneca-Skolkovo Startup Challenge in 2019 and Patents Power 2019.
Ivan Drokin holds four patents related to medical AI and deep learning applications. These include a method for combining patient information through deep learning models into unified representations for diagnostics and prognosis, an invention covering patient data representation and transfer for machine learning algorithm analysis, a comprehensive patent covering the entire process from data acquisition to final machine learning predictions for diagnosing organ pathologies, and an approach based on point-cloud representation of human body CT scans with associated models and technologies.