Projects sharing features with Flower

SecretFlow is a privacy computing framework and platform designed for secure multi-party computation, federated learning, and privacy-preserving data analysis across independent nodes. It provides a management system to coordinate secure workloads and cryptographic tasks across a distributed cluster. The project enables joint data analysis and machine learning on partitioned datasets using cryptographic protocols. It allows for the training of models and the execution of analytical queries across multiple parties without exposing raw source information to any single participant. The framewor

NNI is an AutoML toolkit designed to automate machine learning lifecycles. It functions as a hyperparameter optimization framework, a neural architecture search tool, and a model compression suite. The project provides a distributed training orchestrator to manage machine learning workloads across local machines, remote servers, and cloud platforms. It enables the discovery of efficient model structures through reinforcement learning and one-shot optimization methods, while utilizing Bayesian and evolutionary algorithms to automate hyperparameter tuning. Additional capabilities include tools

This project provides a comprehensive technical guide and framework for engineering large-scale machine learning systems. It covers the full lifecycle of model development, focusing on the infrastructure and computational principles required to build, train, and serve generative AI models across distributed GPU clusters. The repository distinguishes itself by offering deep-dive tutorials and implementation strategies for complex system challenges. It emphasizes high-performance architectural primitives, such as collective communication orchestration, distributed tensor sharding, and static gr

SkyPilot is a multi-cloud AI orchestrator and distributed task scheduler designed to launch and manage AI workloads across various cloud providers, Kubernetes, and Slurm clusters. It functions as an infrastructure-as-code framework that uses declarative files to define resource requirements and setup commands for consistent execution across different environments. The project differentiates itself through automated cost optimization, selecting the most affordable GPU or TPU hardware and managing spot instances to reduce expenses. It also provides a remote development environment that bridges

The Adversarial Robustness Toolbox (ART) is an open-source library that provides a unified framework for evaluating, defending, and certifying machine learning models against adversarial threats. It wraps models from any framework behind a common estimator interface, enabling composable pipelines for attack generation, defense application, robustness certification, and privacy auditing across evasion, poisoning, and extraction threats. The library distinguishes itself by covering the full adversarial ML security lifecycle within a single toolkit. It supports gradient-based adversarial example

This project is a functional programming library and toolkit for building production TypeScript applications. It provides a system for managing concurrency, error handling, and resource lifecycles using functional effects. The project distinguishes itself through a comprehensive suite of specialized toolkits, including a dependency injection framework for decoupling service implementations, a workflow orchestrator for coordinating durable processes, and a SQL database toolkit for consistent data operations across multiple dialects. It also implements an OpenTelemetry instrumentation library f

Osquery is a unified endpoint monitoring framework that exposes operating system internals as relational tables. By representing hardware, network, and process activity as structured data, it allows users to retrieve system state and configuration information using standard SQL syntax. The system distinguishes itself through a cross-platform abstraction layer that normalizes disparate operating system interfaces into a consistent schema across Windows, macOS, and Linux. It supports both interactive local analysis via a command-line shell and distributed fleet orchestration, where recurring qu

GPT2-Chinese is a Chinese language model implementation based on the GPT-2 architecture. It provides a causal language model trainer and a natural language generation tool designed for training and generating human-like Chinese text sequences. The system integrates a BERT tokenizer to process Chinese corpora into manageable units for machine learning. It enables the development of predictive text models that can generate specific patterns, such as news or poetry, through prompt-based text completion. The project covers a full workflow including text tokenization, model training using a trans

FastStream is an asynchronous Python framework designed for building event-driven microservices. It provides a unified abstraction layer for interacting with various message brokers, enabling developers to manage event production and consumption through a consistent interface while maintaining access to native provider-specific features. The framework centers on a decorator-based routing model that binds application logic directly to broker topics, supported by a built-in dependency injection container that resolves resources at runtime. The framework distinguishes itself through its deep int

jetson-inference is a set of libraries and tools for executing optimized deep learning models on embedded GPU hardware. Its primary purpose is to enable real-time computer vision and AI inference at the edge with low latency and high throughput. The project distinguishes itself through high-performance streaming analytics and the ability to execute concurrent AI pipelines on auto-grade silicon. It provides specialized support for multi-sensor stream processing, utilizing zero-copy data transport to load camera frames directly into GPU memory. The codebase covers a broad surface of capabiliti

tarpc is a remote procedure call framework for Rust that generates server traits and client stubs for network communication. It provides an asynchronous service interface that allows remote functions to be called as if they were local asynchronous tasks. The framework utilizes a serialized transport layer to move request and response data via TCP sockets, Unix domain sockets, or in-process channels. It includes a request lifecycle interceptor for injecting custom logic to handle authorization, throttling, and load balancing, as well as a distributed request telemetry system that attaches dead

This project is an end-to-end type-safe API framework designed to synchronize data structures between frontend and backend codebases without the need for manual code generation. By leveraging TypeScript’s type inference, it allows developers to invoke server-side functions directly from the client as if they were local methods. This remote procedure call approach abstracts away the complexities of HTTP verbs and URL structures, streamlining the full-stack development workflow into a unified experience. The framework distinguishes itself through a modular router architecture that organizes bac

MagicOnion is a .NET RPC framework used to build type-safe remote procedure call services using shared interfaces for communication between servers and clients. It functions as a binary serialization framework and a distributed service orchestrator, providing a network layer for low-latency streaming and real-time data exchange. The project distinguishes itself through an AOT-compatible network client that uses source generators to support ahead-of-time compilation in restricted runtime environments. It enables bidirectional communication and real-time data streaming, allowing servers and mul

grpc-go is a Go language implementation of the gRPC framework, providing a remote procedure call library for high-performance service communication. It uses the HTTP/2 protocol to execute functions on remote servers as if they were local methods and utilizes protobuf service bindings to generate type-safe client and server code. The project features a bidirectional streaming transport that supports asynchronous, full-duplex message streams between clients and servers. This networking layer allows for various communication patterns, including client-to-server and server-to-client streaming, to

Connect-go is a library for building type-safe remote procedure call services that are wire-compatible with gRPC and HTTP. It provides a Go implementation of Protocol Buffers for structured data exchange, enabling the creation of multi-protocol RPC servers and generated type-safe clients. The framework is distinguished by its polyglot protocol support, allowing a single server handler to serve requests using several different protocols simultaneously over HTTP. This includes the ability to toggle between different transport protocols for the same service and the use of pluggable content negot

RLinf is a distributed reinforcement learning orchestrator and embodied AI training framework. It provides the infrastructure to train vision-language-action models and robotic policies using a combination of reinforcement learning and supervised fine-tuning. The system is designed for scaling workloads across GPU clusters, managing the placement of actors, rollout workers, and environment components. It features a specialized robotics data collection pipeline for gathering teleoperated demonstrations and simulation trajectories into standardized replay buffers, alongside a hardware interface

Weaver is a distributed application framework and remote procedure call system that allows developers to organize logic into independent components. It provides a multi-process execution environment where these components communicate via automated serialization, enabling applications to run as a single unit locally or as a distributed system across multiple cloud machines. The framework distinguishes itself through a configuration-driven topology mapping that allows the same logic to execute as a local function call or a remote network request without altering the business logic. It includes

This project is a research-oriented platform designed for simulating decentralized machine learning environments. It provides a framework for training models across multiple client nodes while keeping raw data localized, enabling the evaluation of model convergence and performance under various distributed network conditions. The system utilizes a parameter-server architecture to coordinate training, where a central coordinator manages the global model state and aggregates weight updates from distributed participants. By decoupling the training orchestration logic from the underlying neural n

Thrift is a cross-language remote procedure call framework and data serialization protocol. It provides an interface definition language to specify data types and service interfaces in a neutral format, enabling the automated generation of client and server code across multiple programming languages. The project functions as a polyglot service communicator using a layered software stack to ensure interoperable communication. It focuses on implementing cross-language remote procedure calls and transforming complex data structures into standardized formats for efficient network transport. The

brpc is a high-performance C++ RPC framework and network programming library designed for building distributed systems. It functions as a multi-protocol RPC server capable of hosting and detecting multiple communication protocols, including gRPC, Thrift, HTTP, Redis, and Memcached, on a single TCP port. The project distinguishes itself through high-throughput data transport and memory efficiency, utilizing RDMA-based transport to bypass the kernel TCP stack and zero-copy memory management to eliminate data duplication. It also implements the Raft algorithm for consensus-based state replicatio

MagicOnion is a remote procedure call framework for the .NET platform and Unity. It functions as a binary API protocol and real-time communication engine that uses shared interfaces to provide type-safe communication between clients and servers. The framework distinguishes itself through its compatibility with ahead-of-time compilation environments, utilizing source generators for client proxies and serialization to support mobile and console platforms including those using IL2CPP. It provides a specialized middleware layer for Unity that handles engine-specific data types and ensures runtime

This is a remote procedure call framework built on Netty for asynchronous communication between distributed services. It provides the infrastructure necessary to execute remote procedure calls across a network, allowing clients and servers to interact without blocking. The framework utilizes Zookeeper for service registration and dynamic discovery, ensuring that service providers can be located automatically at runtime. It includes a service versioning manager to organize and track different iterations of service interfaces, enabling side-by-side deployments of multiple versions. The system

gRPC Java is a library for building distributed systems that enables client and server applications to interact through remote procedure calls. It uses protocol buffers to define service interfaces and exchange structured data, providing a framework for connecting microservices across different environments. The framework utilizes an asynchronous networking foundation to manage concurrent connections and relies on a multiplexed binary protocol to facilitate efficient data exchange. It supports pluggable transport interceptors, allowing for the injection of custom logic into the request-respon

anytype-ts is a TypeScript client library and offline-first knowledge base client designed for managing structured layouts and modular blocks within a private distributed database. It functions as a zero-knowledge data synchronizer and schema-driven API wrapper that enables the secure replication of encrypted data across devices. The library utilizes peer-to-peer connectivity and zero-knowledge encryption to ensure data remains private and unreadable to hosting infrastructure. It employs a schema-driven approach to compile data definitions into typed language bindings, ensuring consistent com

LearningNotes is a technical knowledge base and engineering study guide focused on Android framework internals, system architecture, and mobile performance optimization. It serves as a reference for analyzing the Android boot sequence, process bootstrapping, and system service initialization. The project provides detailed guides on mobile performance, including strategies for reducing memory footprints, identifying memory leaks, and optimizing image decoding. It further covers Android inter-process communication using AIDL and the Binder kernel driver, as well as software architecture manuals

deepstream.io is an open-source realtime server that synchronizes JSON records, events, and remote procedure calls across clients and backend services. It functions as a realtime data sync server, event pub/sub server, record database server, and RPC server, all within a single platform. The server authenticates and authorizes every message using multiple strategies including JWT, HTTP, and file-based credentials, with a declarative permission language controlling access to records, events, and RPCs at a granular level. The platform distinguishes itself through its combination of realtime dat

Cellnet is a Go networking framework designed for building distributed systems and peer-to-peer networks. It provides a multi-protocol transport layer that standardizes message exchange across TCP, UDP, and WebSockets, utilizing a binary message codec system to map message types to unique identifiers for network transmission. The framework is distinguished by its remote procedure call library, which enables the execution of functions on remote peers using synchronous or asynchronous request patterns. It further differentiates itself with an asynchronous event dispatcher that processes network