Subnet 65

Tao Private Network (TPN)

Metanova Labs

TPN is a decentralized subnet designed to empower secure, private, and geographically diverse internet access through a global network of VPN providers.

SN65 : Tao Private Network (TPN)

SubnetDescriptionCategoryCompany
SN65 : Tao Private Network (TPN)VPNInfrastructure

TPN is a decentralized subnet designed to empower secure, private, and geographically diverse internet access through a global network of VPN providers.s a role in rapidly screening a billion-sized molecular library, optimizing the search for high-affinity, synthesizable drug candidates.

01 Multimodal Data Processing

We collect and process diverse types of open multimodal data to accelerate and enhance AGI research, creating one of the largest open-source multimodal datasets.

02 Future Event Prediction

We enhance the forecasting of future events by identifying patterns and trends in large datasets, empowering individuals to make better-informed decisions.

03 Home Pricing Prediction

We modernize outdated real estate valuation models by developing AI-driven alternatives. Our predictive models for home prices and sales dates aim to establish a more accurate and dynamic pricing system.

04 Detecting AI-generated media

We create advanced models capable of reliably distinguishing between authentic and synthetic content.

05 Data Structuring

We facilitate and incentivize the transformation of unstructured data—such as videos, podcasts, text, and images—into structured formats suitable for AI training.

06 AI Math Solver

We create advanced AI-powered tools designed to tackle a broad spectrum of mathematical problems, ranging from basic arithmetic to advanced calculus.

07 Coding Assistants

We develop AI coding assistants that enhance workflows through intelligent code generation, completion, and interactive Q&A, enabling developers to focus on creativity and problem-solving.

08 Chunking and RAG

We build, host, and deploy intelligent chunking systems that enhance intrachunk similarity and interchunk dissimilarity, improving the efficiency and precision of data retrieval and processing.

09 Context Compression

We accelerate AI inference by compressing long natural language token sequences into shorter, meaningful representations, reducing computational costs and speeding up inference times.

10 Algorithmic Optimization

We address computational challenges such as determining the shortest path, optimizing network flows, or finding optimal arrangements using advanced graph algorithms and optimization techniques.

11 Roleplay LLMs

We develop, fine-tune and integrate roleplay LLMs capable of generating realistic dialogues based on player actions and background, crafting character storylines, designing quests, and producing unique content.

12 Geomagnetic and Environmental Forecasting

We apply advanced machine learning techniques to model and predict geomagnetic disturbances and environmental conditions. Our work enhances forecasting capabilities for space weather events, which impact technological infrastructure, and soil moisture variability, which plays a critical role in climate monitoring and disaster risk management. By integrating real-time data and historical trends, we develop robust predictive solutions that support resilience, sustainability, and informed decision-making.

13 DeFi Yield Optimizers

We create algorithms that strategically allocate assets across predefined pools with varying interest rates, ensuring maximum yield for users.

14 AI Text Detection

We create AI that drives the development of systems that can identify content created by LLMs.

15 AI Object Detection

We develop AI to detect and track objects on images and videos.

Oleg

Peter the Great St. Petersburg Polytechnic University, Applied Mathematics and Programming

Developed a service leveraging an ensemble of LLMs to solve mathematical problems formulated in natural language.

PwC (Data Scientist)

First Line Software (Machine Learning Engineer)

Alex

PhD in Computer Science, University of California

Fine-tuned the LLM “Qwen2.5” for re-ranking tasks in Information Retrieval, achieving significant improvements.

Just AI (Head of NLP R&D)

Kaggle Competitions Master (Gold medals in CV, NLP, and predictive analytics)

Igor

Moscow Institute of Physics and Technology

Developed a scalable system for scraping video content and generating detailed descriptions, forming the basis of a large multimodal dataset.

Sber Devices (Computer Vision Engineer)

Toptal (Senior Data Scientist)

Denis

Bauman Moscow State Technical University (BMSTU)

Built a backtesting system to efficiently test various approaches in information retrieval.

Yandex School of Data Analysis

EPAM Systems (Senior Data Scientist)

Gregory

University of California, Berkeley

Trained a model for multi-class fake detection, achieving superior results compared to competitors.

Master of Engineering in Electrical Engineering and Computer Science (Data Science and Systems track)

Upwork Inc. (Senior Machine Learning Engineer)

MTS AI (Lead Engineer, Head of Data Science and Data Engineering Division)

Michael

Saint Petersburg State Polytechnic University, Physics, Nanotechnology, and Telecommunications (IPNT)

Designed and implemented a microservice architecture for a text-to-tag system, automating scaling and enabling seamless integration of new components.

ABBYY (Head of Group)

University of Oulu (Machine Learning Researcher)

Dmitrii

Moscow Institute of Physics and Technology, Mathematics and Computer Science

Designed a non-ML-based solution for detection of LLM-generated text, significantly enhancing performance.

Pygma.me (Head of AI)

Tinkoff (LLM Inference Engineer / NLP Engineer)