Coding & Development

Fugaku-LLM is a large language model developed by a collaborative team including Tokyo Institute of Technology and RIKEN, trained on Japan's flagship supercomputer, Fugaku. This model boasts enhanced Japanese language abilities, achieving a high score of 5.5 on the Japanese MT-Bench, with a remarkable 9.18 for humanities and social sciences tasks. With 13 billion parameters, it surpasses many other Japanese-developed models. Uniquely, Fugaku-LLM was trained from scratch using proprietary Japanese data, ensuring transparency and safety, unlike models relying on continual learning from existing open models. It is available for both research and commercial use, with its source code on GitHub and the model on Hugging Face, promoting further development and innovative applications in AI.

The Large Generative AI Models in Telecom (GenAINet) is an initiative by the IEEE Communications Society (ComSoc) dedicated to advancing the application of Generative AI within the telecommunications sector. This program focuses on supporting the development of novel architectures specifically tailored for telecom applications, leveraging the capabilities of large Generative AI models. Beyond architectural innovation, GenAINet also plays a crucial role in curating comprehensive dataset libraries essential for training and validating these advanced AI models. Furthermore, the initiative is actively involved in developing sophisticated fine-tuning strategies to optimize Generative AI models for specific telecom challenges. A key objective is to enhance existing communication schemes and protocols, ensuring they are robust and efficient for integration with generative AI agents, thereby pushing the boundaries of what's possible in telecom with AI.

QED Software specializes in designing, developing, and deploying scalable AI/ML systems to address complex business challenges. Their services span strategy and consulting, innovation and IP protection, audit and security, model training and integration, and ongoing monitoring and maintenance. They offer expertise in Agentic AI, Machine Learning, Computer Vision, and Big Data Science, supported by proprietary technologies like InfoFrames for unstructured data, BrightBox for XAI, LITL for active labeling, FeatureBook for ML feature management, and BlueQuail for AI data platforms. QED Software focuses on data-driven AI that scales, ensuring repeatable processes and tangible value for market leaders, with a strong emphasis on scientific rigor and full lifecycle strategic partnerships.

QOUIQ, LLC, operating as "一竞技" (Yi Jing Ji), is a leading platform dedicated to healthy competition, integrating professional e-sports and traditional sports. It provides an AI-powered intelligent matching system that ensures fair play by analyzing player skill and recent performance, eliminating issues like 'bad teammates'. The platform also offers a comprehensive coaching system with former professional players and national-level coaches, providing personalized training plans and over 1000 high-quality instructional videos. Emphasizing healthy competition, it includes a scientific anti-addiction system with daily game time reminders and activity suggestions. QOUIQ hosts hundreds of weekly, monthly, and annual tournaments, offering rewards and opportunities for professional team recommendations. It fosters an active community of over 500,000 users and supports seamless multi-platform access (PC, mobile, tablet) with low-latency cloud-based gameplay.

IDScan.net offers a comprehensive AI-powered identity verification platform designed to build trust and prevent ID fraud. The platform provides powerful cloud-based software, VeriScan, for high-volume ID scanning and identity proofing, enabling businesses to scan IDs, passports, and mDLs. Key features include ID scanning, ID authentication, mobile ID/mDL reading, and remote ID validation. It also specializes in deepfake fraud prevention, DMV API integration, and face matching. Solutions cover ID fraud prevention, access management, and age verification, ensuring compliance and security across various industries. IDScan.net integrates with existing systems through SDKs and APIs, making it easy to embed identity proofing into onboarding or enrollment workflows.

Awesome AI Awesomeness is an extensive, curated list of resources dedicated to artificial intelligence, covering a wide array of topics including Machine Learning, Deep Learning, Computer Vision, and Natural Language Processing. This open-source repository on GitHub provides links to papers, frameworks, datasets, and other research topics, making it an invaluable resource for anyone looking to delve into AI. It's designed to be a comprehensive starting point for AI learning and exploration, offering structured access to various sub-fields and practical applications. The list is continuously updated and welcomes contributions, ensuring its relevance and breadth for researchers, students, and practitioners alike.

Enzai is an enterprise AI governance and enablement platform designed to help organizations manage AI risk and compliance. It provides a centralized system of record for AI systems, models, datasets, and governance decisions, making them auditable and trackable. Unlike generic GRC tools, Enzai is purpose-built for AI, capturing unique metadata of models and autonomous systems. It supports major standards like the EU AI Act, NIST AI RMF, and ISO 42001, ensuring alignment with evolving regulations. The platform offers features for structured intake, comprehensive AI inventory, automated assessments, and real-time oversight, helping to embed governance into everyday AI workflows without hindering innovation.

SagivTech is a leading R&D solutions provider with over 16 years of experience in computer vision, machine learning, deep learning, data science, and code optimization. They develop innovative algorithms for image classification, accelerate applications on mobile platforms, and provide full turnkey solutions. Their expertise includes robust solutions for classification, detection, segmentation, and tracking, seamlessly combining classical computer vision techniques with cutting-edge deep learning. SagivTech also specializes in optimizing code for performance KPIs on NVIDIA GPUs, mobile devices, and traditional CPUs, and delivers high-end data science R&D services to derive actionable insights from complex data. Additionally, they offer professional courses in Image Processing, Computer Vision, Deep Learning, and GPU Computing with CUDA.

Edge Impulse is a comprehensive development platform designed for machine learning on edge devices, supporting everything from microcontrollers to gateways with neural accelerators. It empowers product leaders, AI practitioners, embedded engineers, and OEMs to build smarter edge products by offering capabilities like sensor dataset creation, feature engineering, model optimization, and computer vision algorithms. The platform aims to accelerate time to market, remove complexities in model development, and de-risk deployment with agnostic and scalable edge AI tools. It supports various industries including appliances, health, industrial, and transportation, and offers integrations with popular hardware like Arduino, Qualcomm, and NVIDIA.

Seezo is an AI-powered tool designed to automate and scale security design reviews within existing development workflows. It addresses common challenges like slow turnaround times, inconsistent coverage, and reactive security practices by integrating directly into design documentation and ticketing systems. Seezo maps system components, visualizes data flows, and generates context-specific, explainable security findings. It analyzes design documents and PRDs to produce structured security summaries and threat model artifacts, including components, assets, and trust boundaries. The tool tailors security standards to specific tech stacks and risk profiles, ensuring compliance and internal policy enforcement. Security requirements are delivered as Jira tickets, Slack messages, or Google Doc comments, eliminating context switching for developers. Seezo provides full traceability and audit-ready decision trees for every requirement, showing the exact logic path and rule triggered.

RealNetworks leverages artificial intelligence and machine learning to offer solutions across various domains, including security, communication, and entertainment. Their flagship product, SAFR, provides AI-powered facial recognition specifically designed for security professionals, offering greater visibility and confidence in security operations. Kontxt is another key offering, utilizing AI to secure communication by categorizing A2P messages for mobile carriers, which helps in enhancing customer loyalty and combating spam. Additionally, RealNetworks is known for GameHouse, providing a platform for enjoying games, and RealPlayer, a media player for stunning video experiences. The company focuses on applying AI to real-world problems to enhance security, communication, and digital entertainment.

mindall-e is a PyTorch implementation of a 1.3B text-to-image generation model, trained on 14 million image-text pairs for non-commercial use. This open-source tool, named after minGPT, utilizes a two-stage autoregressive model. It replaces the original DALL-E's Discrete VAE with VQGAN for more effective high-quality sample generation. The model can generate candidate images from a text prompt and re-rank them using OpenAI's CLIP. It also supports transfer learning for class-conditional and unconditional generation tasks, allowing fine-tuning on datasets like ImageNet. The repository provides code snippets for sampling and interactive demos, along with quantitative results demonstrating its performance against other models.

Autoregressive-Models-in-Vision-Survey is an open-source GitHub repository that serves as a comprehensive survey of autoregressive models in the field of computer vision. It curates a list of academic papers, including those on image generation (unconditional, text-to-image, image-to-image translation, image editing), video generation, embodied AI, 3D generation, and multimodal generation. The repository is designed to be a valuable resource for researchers and practitioners, offering insights into the latest advancements and trends in autoregressive visual generation. It includes sections on pixel-wise and token-wise generation, as well as benchmarks, analysis, and evaluation metrics. The project is actively maintained with updates on new research and welcomes contributions from the community.

WitHub is an AI development platform designed to help developers build AI applications from scratch. It provides a comprehensive pipeline and a suite of tools to accelerate experiments and customize workflows, allowing developers to focus on creating valuable AI products rather than managing infrastructure. The platform includes a library of AI tools equipped with capabilities for text processing, translation, and vision tasks, making it a versatile solution for various AI development needs. WitHub aims to streamline the development process, enabling faster iteration and deployment of AI solutions.

Hoop.dev, specifically Rails Guard, is an AI-powered security solution designed for Rails console sessions. It offers live AI data masking to protect sensitive data from being exposed in real-time, ensuring compliance without impacting operational speed. The tool integrates with a single line of code, providing invisible guardrails for Rails console access. Key features include passwordless authentication via Google SSO with MFA, automated employee onboarding and off-boarding, and granular access control based on user groups. Hoop.dev also records all console sessions, offering comprehensive auditability to track who did what, when, and where. It aims to reduce reliance on static keys and credentials for remote access and provides transparent operation with no changes to existing client workflows. Additionally, it helps automate compliance with regulations like HIPAA, SOC 1/2, PCI, and GDPR.

Awesome-GraphRAG is a comprehensive repository offering a curated list of resources focused on graph-based retrieval-augmented generation (GraphRAG). This valuable resource is classified according to a survey on Graph Retrieval-Augmented Generation for Customized Large Language Models, making it easy for researchers to navigate. It continuously updates with the latest surveys, research papers, benchmarks, and open-source projects in the GraphRAG domain. The repository details the evolution of GraphRAG, comparing it with traditional RAG and highlighting its innovations in knowledge representation, retrieval mechanisms, and search algorithms. It serves as an essential hub for academics and practitioners interested in advancing LLM applications through graph-structured knowledge.

adaptAI Inc. specializes in guiding mid-sized businesses through the complex landscape of AI implementation, emphasizing governance and business value. They offer services such as AI Governance, LLM Hosting/Private Models, AI Agents, Fractional Data Science, and AI MVP Product Development. The company aims to make AI technology accessible and integrated into clients' operations, ensuring responsible development and use. adaptAI provides solutions for challenges like manual task tracking, complex data analysis, and enhancing enterprise search with AI-powered features, leading to increased efficiency, reduced costs, and improved decision-making.

AI Person ID offers a universal digital identity solution for both humans and artificial intelligences, designed to foster secure coexistence, transactions, and communication. It provides a unique ID with public keys, an inbox and email for verified communication, and a financial account for managing payments and crypto. The platform emphasizes interoperability across various AI systems and platforms, aiming to integrate finance, communications, and physical presence. Its core purpose is to create a sovereign and interoperable identity that belongs to the individual, not a company or government, bringing trust back to the age of machines.

2021.AI provides the GRACE AI Platform, an Enterprise AI solution designed to help organizations efficiently and responsibly manage, scale, and govern their AI operations. The platform ensures sovereignty, resilience, and robustness, giving users full control and oversight of their AI. Key modules include GenAI for safe and secure optimization, MLOps for efficient development and deployment, and Governance to ensure compliance and manage risks. It also features an AI Registry for comprehensive AI portfolio management and solutions for safeguarding and monitoring LLMs. 2021.AI assists organizations from concept to execution, ensuring secure, compliant, and future-proof AI implementation.

Big Vision offers world-class AI consulting services with deep expertise in building and deploying end-to-end solutions across AI, Computer Vision, and Deep Learning. They partner with businesses to create specific AI-driven solutions, establish intelligent processes, and implement best practices to reduce human effort and enable faster decision-making. Their enterprise-grade solutions currently serve millions of requests monthly in critical production environments. Big Vision emphasizes smart data research, world-class implementation, and collaborative deployment using MLOps best practices. They also focus on responsible use of technology, rigorous ethical review, and continuous support for long-term client success. Beyond consulting, Big Vision is committed to empowering a global AI workforce through acclaimed AI courses and expert blogs.

Step-Audio-EditX is an open-source, 3B-parameter LLM-based Reinforcement Learning audio model designed for expressive and iterative audio editing. It specializes in modifying emotion, speaking style, and paralinguistic features within audio. The tool also provides robust zero-shot text-to-speech (TTS) capabilities, supporting languages like Mandarin, English, Sichuanese, Cantonese, Japanese, and Korean. Users can control polyphonic pronunciation, choose from dozens of emotion and speaking style options, and precisely manage 10 types of paralinguistic features for more natural synthetic audio. The project provides training and inference code, model weights, and a demo page for exploration.

Tree-of-thought-prompting is an open-source project on GitHub that introduces a novel Tree-of-Thought (ToT) Prompting technique to significantly boost the reasoning capabilities of Large Language Models (LLMs) like ChatGPT. Building upon the established Chain-of-Thought (CoT) prompting, ToT Prompting allows LLMs to explore multiple reasoning paths, self-evaluate, and correct errors autonomously. This method has shown to enable ChatGPT 3.5 to correctly answer complex questions that previously only ChatGPT 4 could handle. The project provides example prompts and encourages community contributions to refine and expand the technique, offering a single-prompt solution for enhanced LLM performance without requiring multiple API calls.

Reflection AI is dedicated to building and making accessible frontier open intelligence. The company focuses on advancing AI research and development, with a team experienced in creating large language models (LLMs) at leading AI labs such as DeepMind, OpenAI, and Anthropic. Their mission is to democratize access to cutting-edge AI, enabling broader participation in and benefit from advanced intelligence. They invite collaboration through partnerships and offer career opportunities for those interested in contributing to their research.

QuantumCoreX delivers advanced AI, computing, and and cybersecurity solutions, empowering businesses with innovative technology to enhance efficiency, security, and growth in today’s digital landscape. They specialize in integrating AI into organizations, optimizing data for AI applications, and enhancing software and mobile apps with cutting-edge AI. Their services include Artificial Intelligence, Data Engineering, Generative AI, DevOps, and Development Services. QuantumCoreX also provides custom cybersecurity solutions, including on-demand services, red teaming, penetration testing, and defense strategies to disrupt adversaries and protect businesses from emerging threats. They serve diverse industries such as Healthcare & LifeSciences, Technology & SaaS, Manufacturing & Logistics, Finance & Insurance, and Retail & eCommerce.