Akash Singh

magenta-realtime-2-iphone. Google's MRT2 → Core ML, full Apple Neural Engine execution, zero GPU. Runs locally on iPhone without burning your hand.

magenta-realtime. Magenta RealTime 2: An Open-Weights Live Music Model

avtr-1. AVTR-1: Avatars that listen back

stable-audio-3. Python

xybrid. Build apps powered by on-device AI

ds4. DeepSeek 4 Flash and PRO local inference engine for Metal, CUDA and ROCm

MTPLX. 2.24x decode TPS increase On Qwen 3.6 27B @ temp 0.6 | Native MTP Speculative Decoding On Apple Silicon With No External Drafter.

LocalVQE. Lean neural real-time acoustic echo cancellation with soft delay estimation - GGML and PyTorch inference

MaAI. A real-time software for turn-taking, backchannel, and head-nodding prediction

mlx-lm-lora. Train Large Language Models on MLX.

lolcats. Repo for "LoLCATs: On Low-Rank Linearizing of Large Language Models"

macmon. 🦀🌡️ Real-time system monitor for Apple Silicon Macs (M1–M5). No sudo. TUI, JSON/Prometheus metrics server, and Rust library.

RuView. π RuView turns commodity WiFi signals into real-time spatial intelligence, vital sign monitoring, and presence detection — all without a single pixel of video.

kitten_tts_rs. Rust implementation of KittenTTS. CLI and API server.

tada. Open Source Speech Language Model

simply. Minimal and scalable research codebase in JAX, designed for rapid iteration on frontier research in LLM and other autoregressive models.

ltx2-castlehill. CastleHill: Separable Causal Diffusion / Varitaion Flow Maps for LTX-2 long-form video generation

Helios. Helios: Real Real-Time Long Video Generation Model

runanywhere-sdks. Production ready toolkit to run AI locally

ANE. Training neural networks on Apple Neural Engine via reverse-engineered private APIs

zeroclaw. Fast, small, and fully autonomous AI personal assistant infrastructure, any OS, any platform — deploy anywhere, swap anything 🦀

qwen3_tts_rs. A Rust implementation of the Qwen3 Text-to-Speech (TTS) model inference.

faster-qwen3-tts. Real-time text-to-speech with Qwen3-TTS

openfang. Open-source Agent Operating System

python-apple-fm-sdk. Python bindings for access to the on-device model at the core of Apple Intelligence through the Foundation Models framework

personaplex-mlx. PersonaPlex on Apple Silicon: an MLX port of NVIDIA’s full-duplex speech-to-speech model with realtime local/web modes and offline WAV inference.

personaplex. PersonaPlex code.

pocket-tts. A TTS that fits in your CPU (and pocket)

voicebox. The open-source AI voice studio. Clone, dictate, create.

mini-sglang. A compact implementation of SGLang, designed to demystify the complexities of modern LLM serving systems.

ACE-Step-1.5. The most powerful local music generation model that outperforms almost all commercial alternatives, supporting Mac, AMD, Intel, and CUDA devices.

transformerlab-app. The open source research environment for AI researchers to seamlessly train, evaluate, and scale models from local hardware to GPU clusters.

ART. Agent Reinforcement Trainer: train multi-step agents for real-world tasks using GRPO. Give your agents on-the-job training. Reinforcement learning for Qwen3.6, GPT-OSS, Llama, and more!

ml-clara. Python

Open-AutoGLM. An Open Phone Agent Model & Framework. Unlocking the AI Phone for Everyone

vllm-omni. A framework for efficient model inference with omni-modality models

ten-framework. Open-source framework for conversational voice AI agents

VibeVoice. Open-Source Frontier Voice AI

supertonic. Lightning-Fast, On-Device, Multilingual TTS — running natively via ONNX.

llm-compressor. Transformers-compatible library for applying various compression algorithms to LLMs for optimized deployment with vLLM

omnilingual-asr. Omnilingual ASR Open-Source Multilingual SpeechRecognition for 1600+ Languages

TheWhisper. Optimized Whisper models for streaming and on-device use

edge. On-device intelligence.

mamba.py. A simple and efficient Mamba implementation in pure PyTorch and MLX.

higgs-audio. Text-audio foundation model from Boson AI

TinyRecursiveModels. Python

InferenceX. Open Source Continuous Inference Benchmark Research Platform — Kimi K2.7-Code, MiniMax M3, DeepSeekv4, GLM5 - GB200 NVL72 vs MI355X vs B200 vs GB300 NVL72 & soon™ TPUv6e/v7/Trainium2/3 | 开源持续推理基准研究平台 — Kimi K2.7-Code、MiniMax M3、DeepSeekv4、GLM5 - GB200 NVL72 vs MI355X vs B200 vs GB300 NVL72，即将推出™ TPUv6e/v7/Trainium2/3

neutts. On-device TTS model by Neuphonic

liquid-audio. Liquid Audio - Speech-to-Speech audio models by Liquid AI

Qwen3-Omni. Qwen3-omni is a natively end-to-end, omni-modal LLM developed by the Qwen team at Alibaba Cloud, capable of understanding text, audio, images, and video, as well as generating speech in real time.

voxtream. VoXtream is a Full-Stream Zero-shot TTS model with Extremely Low Latency and Speaking rate Control

FluidAudio. Frontier CoreML audio models in your apps — text-to-speech, speech-to-text, voice activity detection, and speaker diarization. In Swift, powered by SOTA open source.

ultravox. A fast multimodal LLM for real-time voice

jan. Jan is an open source alternative to ChatGPT that runs 100% offline on your computer.

LMCache. LMCache: Supercharge Your LLM with the Fastest KV Cache Layer

cocoindex. Incremental engine for long horizon agents 🌟 Star if you like it!

embedding-atlas. Embedding Atlas is a tool that provides interactive visualizations for large embeddings. It allows you to visualize, cross-filter, and search embeddings and metadata.

KittenTTS. State-of-the-art TTS model under 25MB 😻

harmony. Renderer for the harmony response format to be used with gpt-oss

audio-flamingo. PyTorch implementation of Audio Flamingo: Series of Advanced Audio Understanding Language Models

uzu. A high-performance inference engine for AI models

ml-fastvlm. This repository contains the official implementation of "FastVLM: Efficient Vision Encoding for Vision Language Models" - CVPR 2025

runagent. RunAgent simplifies serverless deployment of your AI agents. With a powerful CLI, multi-language SDK support, built-in agent invocation & streaming suppprt.

gallery. A gallery that showcases on-device ML/GenAI use cases and allows people to try and use models locally.

torch-profiling-tutorial. Python

mlx-vlm. MLX-VLM is a package for inference and fine-tuning of Vision Language Models (VLMs) on your Mac using MLX.

unmute. Make text LLMs listen and speak

vui. Real-time voice assistant — WebRTC streaming, faster-whisper ASR, local LLM, Vui Nano (300M) TTS. OpenAI Realtime API compatible. Voice cloning, barge-in, ~9× realtime on a 4090. Apache 2.0.

gemini-fullstack-langgraph-quickstart. Get started with building Fullstack Agents using Gemini 2.5 and LangGraph

chatterbox. SoTA open-source TTS

NLWeb. Main reference implementation for NLWeb, implemented in Python.

ten-vad. Voice Activity Detector (VAD) : low-latency, high-performance and lightweight

awesome-mcp-servers. A collection of MCP servers.

12-factor-agents. What are the principles we can use to build LLM-powered software that is actually good enough to put in the hands of production customers?

mobilerun. Automate your mobile devices with natural language commands - an LLM agnostic mobile Agent 🤖

suna. The Company AI Command Center

dia. A TTS model capable of generating ultra-realistic dialogue in one pass.

Second-Me. Train your AI self, amplify you, bridge the world

system-prompts-and-models-of-ai-tools. FULL Augment Code, Claude Code, Cluely, CodeBuddy, Comet, Cursor, Devin AI, Junie, Kiro, Leap.new, Lovable, Manus, NotionAI, Orchids.app, Perplexity, Poke, Qoder, Replit, Same.dev, Trae, Traycer AI, VSCode Agent, Warp.dev, Windsurf, Xcode, Z.ai Code, Dia & v0. (And other Open Sourced) System Prompts, Internal Tools & AI Models

WorkflowAI. WorkflowAI is an open-source platform where product and engineering teams 
collaborate to build and iterate on AI features.

csm-mlx. An implementation of the CSM(Conversation Speech Model) for Apple Silicon using MLX.

NeMo-Agent-Toolkit. The NVIDIA NeMo Agent toolkit is an open-source library for efficiently connecting and optimizing teams of AI agents.

moshi-finetune. Python

awesome-LangGraph. An index of the LangChain + LangGraph ecosystem: concepts, projects, tools, templates, and guides for LLM & multi-agent apps.

curator. Synthetic data curation for post-training and structured data extraction

mem0. Universal memory layer for AI Agents

namo-sslm. NAMO Small Speech Language Model

HuggingSnap. SmolVLM2 Demo

Orpheus-TTS. Towards Human-Sounding Speech

data-formulator. 🪄 Data Formulator is an interactive AI-powered data analysis system makes it easy to connect, explore and visualize data.

dynamo. A Datacenter Scale Distributed Inference Serving Framework

smoltts. Open TTS models, built for streaming on the edge

Agent-S. Agent S: an open agentic framework that uses computers like a human

openai-agents-python. A lightweight, powerful framework for multi-agent workflows

agentops. Python SDK for AI agent monitoring, LLM cost tracking, benchmarking, and more. Integrates with most LLMs and agent frameworks including CrewAI, Agno, OpenAI Agents SDK, Langchain, Autogen, AG2, and CamelAI

owl. 🦉 OWL: Optimized Workforce Learning for General Multi-Agent Assistance in Real-World Task Automation

awesome-cursorrules. 📄 Configuration files that enhance Cursor AI editor experience with custom rules and behaviors

csm. A Conversational Speech Generation Model

mlx-audio. A text-to-speech (TTS), speech-to-text (STT) and speech-to-speech (STS) library built on Apple's MLX framework, providing efficient speech analysis on Apple Silicon.

moonshine. Very low latency speech to text, intent recognition, and text to speech, for building voice agents and interfaces

Awesome-LLM-Post-training. Awesome Reasoning LLM Tutorial/Survey/Guide

ReCall. ReSearch: Learning to Reason with Search for LLMs via Reinforcement Learning & ReCall: Learning to Reason with Tool Call for LLMs via Reinforcement Learning

hyperswitch. Open source, composable payments platform | PCI compliant | SaaS and Self-host options | Enables connectivity to multiple payment, payout, fraud, vault and tokenization providers | Uplifts authorization with intelligent routing and revenue recovery | Reduce payment processing costs with cost observability | Reduces payment ops with reconciliation

Zonos. Zonos-v0.1 is a leading open-weight text-to-speech model trained on more than 200k hours of varied multilingual speech, delivering expressiveness and quality on par with—or even surpassing—top TTS providers.

kokoro. https://hf.co/hexgrad/Kokoro-82M

Moonlight. Muon is Scalable for LLM Training

pipecat. Open Source framework for voice and multimodal conversational AI

fastrtc. The python library for real-time communication

DeepGEMM. DeepGEMM: clean and efficient BLAS kernel library on GPU

ggwave. Tiny data-over-sound library

servers. Model Context Protocol Servers

hibiki. Hibiki is a model for streaming speech translation (also known as simultaneous translation). Unlike offline translation—where one waits for the end of the source utterance to start translating--- Hibiki adapts its flow to accumulate just enough context to produce a correct translation in real-time, chunk by chunk.

OLMoE.swift. Swift

awesome-ai-agents. A list of AI autonomous agents

audio-ai-hub. The hub for audio AI research: papers, open models, benchmarks & datasets across audio LLMs, speech recognition, TTS, music & audio generation.

s1. s1: Simple test-time scaling

R1-V. Witness the aha moment of VLM with less than $3.

open-thoughts. Fully open data curation for reasoning models

ogx. Open GenAI Stack

web-llm. High-performance In-browser LLM Inference Engine

CAG. Cache-Augmented Generation: A Simple, Efficient Alternative to RAG

minimind. 🧠「大模型」2小时完全从0训练64M的小参数LLM！Train a 64M-parameter LLM from scratch in just 2h!

hallo3. [CVPR 2025] Hallo3: Highly Dynamic and Realistic Portrait Image Animation with Video Diffusion Transformer

Matcha-TTS. [ICASSP 2024] 🍵 Matcha-TTS: A fast TTS architecture with conditional flow matching

VoiceStreamAI. Near-Realtime audio transcription using self-hosted Whisper and WebSocket in Python/JS

diffusion-rs. Blazingly fast inference of diffusion models.

cosmos. NVIDIA Cosmos is an open platform of world models, datasets, and tools that enables developers to build Physical AI for robots, autonomous vehicles, smart infrastructure, and more.

open-computer-use. AI computer use powered by open source LLMs and E2B Desktop Sandbox

DeepSeek-V3. Python

ik_llama.cpp. llama.cpp fork with additional SOTA quants and improved performance

large_concept_model. Large Concept Models: Language modeling in a sentence representation space

smolagents. 🤗 smolagents: a barebones library for agents that think in code.

acwj. A Compiler Writing Journey

pytorch_geometric. Graph Neural Network Library for PyTorch

taichi. Productive, portable, and performant GPU programming in Python.

fiftyone. Refine high-quality datasets and visual AI models

3D-Speaker. A Repository for Single- and Multi-modal Speaker Verification, Speaker Recognition and Speaker Diarization

system-design-primer. Learn how to design large-scale systems. Prep for the system design interview. Includes Anki flashcards.

tldraw. Build infinite canvas apps in React with the tldraw SDK. World's best, top-most agent recommended #1 five star SDK.

WonderWorld. Code release for https://kovenyu.com/WonderWorld/

monolith. A Lightweight Recommendation System

genesis-world. Simulation platform for general-purpose robotics & embodied AI learning.

onyx. Open Source AI Platform - AI Chat with advanced features that works with every LLM

search-and-learn. Recipes to scale inference-time compute of open models

CosyVoice. Multi-lingual large voice generation model, providing inference, training and deployment full-stack ability.

synthetic-data-generator. Build datasets using natural language

metamotivo. The first behavioral foundation model to control a virtual physics-based humanoid agent for a wide range of whole-body tasks.

gazelle. Gaze-LLE: Gaze Target Estimation via Large-Scale Learned Encoders (CVPR 2025, Highlight)

rust-music-theory. A music theory guide written in Rust.

papers-we-love. Papers from the computer science community to read and discuss.

MMAudio. [CVPR 2025] MMAudio: Taming Multimodal Joint Training for High-Quality Video-to-Audio Synthesis

vocode-core. 🤖 Build voice-based LLM agents. Modular + open source.

WavTokenizer. [ICLR 2025] SOTA discrete acoustic codec models with 40/75 tokens per second for audio language modeling

evo-memory. Code to train and evaluate Neural Attention Memory Models to obtain universally-applicable memory systems for transformers.

spdl. Scalable and Performant Data Loading

TRELLIS. Official repo for paper "Structured 3D Latents for Scalable and Versatile 3D Generation" (CVPR'25 Spotlight).

FunASR. Industrial-grade speech recognition toolkit: 170x realtime, 50+ languages, speaker diarization, emotion detection, streaming, and OpenAI-compatible API.

meshgen. Use AI Agents directly in Blender.

isaac_berkeley_humanoid. Python

ClearerVoice-Studio. An AI-Powered Speech Processing Toolkit and Open Source SOTA Pretrained Models, Supporting Speech Enhancement, Separation, and Target Speaker Extraction, etc.

filament. Filament is a real-time physically based rendering engine for Android, iOS, Windows, Linux, macOS, and WebGL2

weathernext. Python

ProtoMotions. ProtoMotions is a GPU-accelerated simulation and learning framework for training physically simulated digital humans and humanoid robots.

rig. ⚙️🦀 Build modular and scalable LLM Applications in Rust

activeft. PyTorch library for Active Fine-Tuning

smollm. Everything about the SmolLM and SmolVLM family of models

llguidance. Super-fast Structured Outputs

rust-gpu. 🐉 Making Rust a first-class language and ecosystem for GPU shaders 🚧

moshi-swift. Swift

rust-chat-app. Rust

ShowUI. [CVPR 2025] Open-source, End-to-end, Vision-Language-Action model for GUI Agent & Computer Use.

RoBart. An iPhone-based LLM-controlled autonomous robot.

humanoid-bench. Python

copper-rs. Copper is an operating system for robots - build, run, and replay your entire robot deterministically.

pinocchio. A fast and flexible implementation of Rigid Body Dynamics algorithms and their analytical derivatives

loco-mujoco. Imitation learning benchmark focusing on complex locomotion tasks using MuJoCo.

vad-rs. Speech detection using silero vad in Rust

mochi. The best OSS video generation models, created by Genmo

DeepFilterNet. Noise supression using deep filtering

rl-tools. The Fastest Deep Reinforcement Learning Library

Gymnasium. A standard API for single-agent reinforcement learning environments, with popular reference environments and related utilities (formerly Gym)

egui. egui: an easy-to-use immediate mode GUI in Rust that runs on both web and native

SageAttention. [ICLR2025, ICML2025, NeurIPS2025 Spotlight] Quantized Attention achieves speedup of 2-5x compared to FlashAttention, without losing end-to-end metrics across language, image, and video models.

mini-omni. open-source multimodal large language model that can hear, talk while thinking. Featuring real-time end-to-end speech input and streaming audio output conversational capabilities.

chaos. James Gleick's Chaos: The Software

asm_rs_bench. gemv, gemm, llama.cpp arm64 optimised bench

zeroth-bot. 3D-printed open-source humanoid robot platform for sim-to-real and RL

Kinetix. Reinforcement learning on general 2D physics environments in JAX. ICLR 2025 Oral.

open-oasis. Inference script for Oasis 500M

PhysX. NVIDIA PhysX SDK

mujoco. Multi-Joint dynamics with Contact. A general purpose physics simulator.

Gaia. Gaia Physics Engine

bevy. A refreshingly simple data-driven game engine built in Rust

Cosmos-Tokenizer. A suite of image and video neural tokenizers

awesome-o1. A bibliography and survey of the papers surrounding o1

bcr_bot. A mobile robot simulation with differential drive, depth camera, IMU and 2D LiDAR that supports ROS, ROS2, Gazebo, Ignition, and ISAAC Sim

mg2hfbot. Converts MimicGen dataset into LeRobot format, to train and evaluate the ACT, BC, and diffusion policies

ddia-references. Literature references for “Designing Data-Intensive Applications”

ZathuraDbg. An emulation based tool for learning and debugging assembly.

rathole. A lightweight and high-performance reverse proxy for NAT traversal, written in Rust. An alternative to frp and ngrok.