AGENTSIM is a computational platform for large-scale agent-based modelling, supporting the analysis of complex systems and emergent behaviours across domains such as public policy, urban planning and social dynamics.

CHAINTEL is a GPU-accelerated testbed designed for large-scale analysis of Bitcoin and Ethereum data using advanced graph analytics and machine learning. It provides a high-capacity environment for researchers to identify entities, trace transaction flows, and study the structural evolution of public blockchains.

DEEPLAB is a high-performance research infrastructure designed for the full lifecycle of deep learning models. It supports scalable training and benchmarking for diverse architectures, from CNNs to transformers, across applications like predictive analytics and signal processing.

EVOLAB is a GPU-accelerated environment specialized for evolutionary computation and genetic programming, utilizing NVIDIA H100 and RTX 6000 hardware to handle massive population-level parallelism. 

GRAPHNET is a high-performance platform for modeling and visualizing complex networked systems, leveraging GPU-accelerated libraries and graph databases to study large-scale structural dynamics. By integrating classical network science with modern AI, it supports the full lifecycle of graph analytics across domains like cybersecurity, fraud detection, and social network analysis.

MEDVISION is a high-performance platform specifically engineered to handle the intense memory demands of medical image processing and deep learning. It supports the full research lifecycle, including segmentation, anomaly detection, and computer-aided diagnosis, across various clinical imaging modalities and standard data formats.

Combining multi-core CPUs with GPU acceleration, the MMLAB platform enables the scalable development and benchmarking of both classical and modern ML models. It serves as a versatile environment for building industry-critical applications such as predictive analytics, anomaly detection, and intelligent automation systems.

LLM-Train is dedicated to the local development and customization of LLM without external API reliance. It supports the full model lifecycle, from pre-training to advanced fine-tuning techniques like LoRA and QLoRA, for models such as Llama and Mistral. The platform is optimized for creating domain-specific applications in areas like technical documentation and specialized knowledge bases.