Ahsan Umar_

Home / Entities

Entities

People, organizations, labs, and tools referenced across this site.

  • › American Innovators Network American Innovators Network (AIN) is a policy advocacy network that organized and led the coalition of 35+ signatories for the "Open Weights and American AI Leadership" position paper (July 2026). The network focuses on maintaining American technological competitiveness through open innovation ecosy Jul 24, 2026
  • › Andreessen Horowitz (a16z) Andreessen Horowitz (a16z) is a prominent Silicon Valley venture capital firm that signed the "Open Weights and American AI Leadership" coalition statement (July 2026). The firm has been a vocal advocate for open-source AI and open-weight models as essential infrastructure for American competitivene Jul 24, 2026
  • › Hugging Face Hugging Face is the **central platform and community hub for open-weight models**. As a signatory of the Open Weights coalition, HF represents the **distribution layer** — the infrastructure that makes open weights discoverable, runnable, and composable. | Metric | Scale | |--------|-------| | Model Jul 24, 2026
  • › Linux Foundation The Linux Foundation (LF) is the **world's largest open source steward**, hosting 900+ projects including Linux, Kubernetes, PyTorch, Node.js, and now AI frameworks. As a signatory, LF brings **governance expertise** — how to run open projects that are neutral, sustainable, and industry-aligned. | P Jul 24, 2026
  • › Meta (Facebook) Meta Platforms (NASDAQ: META) is the world's leading developer of open-weight foundation models through its **Llama model family**. As a signatory to the Open Weights coalition, Meta advocates for open-weight models as the foundation for AI innovation and American competitiveness. | Model | Release Jul 24, 2026
  • › Microsoft Microsoft Corporation (NASDAQ: MSFT) is a multinational technology company headquartered in Redmond, Washington. As the publisher and lead signatory of the "Open Weights and American AI Leadership" position paper (July 2026), Microsoft positions itself as a champion of open-weight AI models as essen Jul 24, 2026
  • › Mistral AI Mistral AI is a Paris-based AI lab founded in 2023 by former DeepMind and Meta researchers. It is the **leading European open-weight model developer** and a signatory to the Open Weights coalition, advocating for open models as essential for both American and European AI sovereignty. | Model | Relea Jul 24, 2026
  • › NVIDIA NVIDIA (NASDAQ: NVDA) is the dominant provider of AI compute infrastructure (GPUs, networking, software stack). As a signatory of the Open Weights coalition, NVIDIA's position reflects its business interest in **expanding the total addressable market for AI compute** — open weights drive more traini Jul 24, 2026
  • › Y Combinator Y Combinator (YC) is the world's most prominent **startup accelerator**, having funded 5,000+ companies including Airbnb, Stripe, Dropbox, Coinbase, and hundreds of AI startups. As a signatory to the Open Weights coalition, YC represents the **startup ecosystem's interest** in open weights as essent Jul 24, 2026
  • › Inkling Thinking Machines Lab's 975B total / 41B active parameter open-weight multimodal MoE model (Apache 2.0) with hybrid local/global attention, short convolutions, embedding RMSNorm, and learned relative-position bias instead of RoPE. 1M token context, supports BF16/MXFP8/NVFP4. Jul 19, 2026
  • › Kimi K3 Moonshot AI's 2.8T parameter open-weight flagship model — first in the 3-trillion class — built on Kimi Delta Attention (KDA), Attention Residuals (AttnRes), and Stable LatentMoE (16/896 experts), with 1M context window and native multimodality. Jul 19, 2026
  • › LeWorldModel (LeWM) LeWM is an end-to-end JEPA world model that learns directly from pixels using only a prediction loss + SIGReg regularizer — no pretrained visual features (unlike DINO-WM), no reconstruction decoder, no temporal smoothness loss. Achieves 96% success on Push-T with ViT-S predictor, significantly outperforming PLDM (78%) and DINO-WM (80.67%). Jul 19, 2026
  • › Moonshot AI Chinese AI research lab (Moonshot AI) developing the Kimi series of frontier open-source LLMs, known for maintaining open-source parameter scale leadership and pioneering Kimi Delta Attention, Attention Residuals, and Stable LatentMoE architectures. Jul 19, 2026
  • › Thinking Machines Lab AI research lab (Thinking Machines Lab, Inc.) founded by Mira Murati, developing open-weight multimodal models (Inkling) and the Tinker fine-tuning/platform customization service. Jul 19, 2026
  • › VL-JEPA VL-JEPA extends the JEPA paradigm to vision-language, learning a joint image-text embedding space using only SIGReg regularization — no contrastive loss (CLIP), no generative loss (LLaVA/Flamingo), no reconstruction (MAE). Achieves competitive zero-shot classification, retrieval, and VQA. Jul 19, 2026
  • › Claude Code Claude Code is Anthropic's command-line interface for software engineering powered by the Agent SDK. It exposes the same 5-step autonomous loop (receive → evaluate → execute → feedback → repeat) with production-grade controls: turn limits, cost budgets, permission modes, session continuity, and streaming. Jul 16, 2026
  • › Codex (OpenAI) OpenAI Codex is an autonomous coding agent that introduced **Goals** — persistent, thread-scoped objectives that define "what done looks like" and iterate until evidence confirms completion. Goals transform Codex from a prompt-response tool into an evidence-driven autonomous agent. Jul 16, 2026
  • › DreamerV3 DreamerV3 (Hafner et al., 2023) is a **model-based RL agent** that learns a **world model in latent space** and performs **actor-critic planning** entirely in latent space. It achieves **human-level Atari** with **fixed hyperparameters** across domains — a major step toward general world models. Jul 16, 2026
  • › Genie (DeepMind Video World Model) Genie (Bruce et al., DeepMind, 2024) is an **11B parameter video world model** trained on **unlabeled Internet video** (no action labels). It learns **latent actions** from pixels, enabling controllable video generation and zero-shot policy learning — a key breakthrough for learning from passive observation. Jul 16, 2026
  • › Sapient Intelligence Sapient Intelligence is an AI research lab (founded by HRM/TRM authors) focused on **hierarchical and recursive reasoning architectures**. Key researchers include Alexia Jolicoeur-Martineau (lead author of HRM and TRM). The lab collaborates with Samsung SAIL Montréal. Jul 16, 2026
  • › UniSim (DeepMind Universal Simulator) UniSim (Yang et al., DeepMind, 2024) is a **universal neural simulator** trained on **1.6M real-world trajectories** across diverse robots, tasks, and environments. Policies trained in UniSim transfer to real robots with **minimal fine-tuning** — solving sim-to-real via massive trajectory diversity. Jul 16, 2026
  • › Alexia Jolicoeur-Martineau Research Scientist at Samsung SAIL Montréal. Lead author of TRM (Tiny Recursive Model) and co-author of HRM (Hierarchical Reasoning Model). PhD from Mila/Université de Montréal. Research focus: recursive reasoning, deep supervision, ACT, small-model reasoning for ARC-AGI. Emerging star in efficient reasoning architectures. Jul 15, 2026
  • › AlphaTechLogics Applied AI startup/lab where you work as AI Engineer. Focus: computer vision, edge AI, efficient Vision Transformers for deployment. Product-focused, deployment-oriented. Jul 15, 2026
  • › BERT BERT (Bidirectional Encoder Representations from Transformers) — encoder-only Transformer pretrained with masked language modeling (MLM) and next sentence prediction (NSP). Foundation for modern NLP; spawned RoBERTa, DeBERTa, ELECTRA, etc. Jul 15, 2026
  • › DeepSeek AI Chinese AI lab (Hangzhou) producing frontier open-weight LLMs. Known for architectural efficiency innovations: MLA (Multi-head Latent Attention), DSA (Dynamic Sparse Attention), CSA/HCA hybrid attention, mHC (Manifold-Constrained Hyper-Connections), Muon optimizer, FP4/FP8 training, MoE. DeepSeek-V3.2 (128K) → V4 (1M context) push efficient long-context. Jul 15, 2026
  • › Google Research (Brain/DeepMind) Combined Google Brain + DeepMind — the world's leading industrial AI research lab. Originators of Transformer, BERT, T5, PaLM, Gemini, AlphaFold, Diffusion models. Top-tier PhD destination. Jul 15, 2026
  • › GPT-3 GPT-3 (Generative Pre-trained Transformer 3) — 175B parameter decoder-only Transformer. Demonstrated **in-context learning** (few-shot) without gradient updates. Validated scaling laws: performance scales smoothly with compute, params, data. Jul 15, 2026
  • › HuggingFace Transformers The de facto standard Python library for Transformer models. Provides model architectures, training utilities (Trainer), PEFT (LoRA/QLoRA), TRL (RLHF), and seamless HuggingFace Hub integration. Jul 15, 2026
  • › LinkedIn LinkedIn is the primary professional distribution platform for AI/ML technical content. 2024–2025 specs: 3,000 char post limit (125K for Articles), double line breaks critical, Saves drive 5× reach of likes (Hootsuite 2026), semantic matching > keywords, Carousels/PDFs highest reach format, optimal posting 4–6 AM UTC Tue–Thu weekly. Jul 15, 2026
  • › Modal Serverless GPU platform for ML workloads. Spin up A100/H100 in seconds, Python-native decorators, pay-per-second billing. Ideal for training, fine-tuning, batch inference, and episodic compute. Jul 15, 2026
  • › Samsung SAIL Montréal Samsung AI Lab in Montréal (SAIL = Samsung AI Lab). Produced HRM (Hierarchical Reasoning Model) and TRM (Tiny Recursive Model) — recursive reasoning architectures achieving SOTA on ARC-AGI with tiny parameter counts (7M-27M) via deep supervision + ACT. Key researcher: Alexia Jolicoeur-Martineau. Jul 15, 2026
  • › Sebastian Raschka Staff Research Engineer at Lightning AI. Author of "Python Machine Learning," "Machine Learning with PyTorch and Scikit-Learn," "Build a Large Language Model (from Scratch)." PhD from Michigan State University. Writes exceptional technical deep-dives (blog, Substack). Published detailed explainer on DeepSeek DSA. Strong educator + practitioner — potential PhD advisor or industry mentor. Jul 15, 2026
  • › Noam Shazeer Co-author of "Attention Is All You Need"; proposed scaled dot-product attention and multi-head attention. Pioneer of Mixture-of-Experts (MoE) for LLMs. Co-founder of Character.AI (acquired by Google 2024). Jul 15, 2026
  • › Smart Technology Lab University computer vision lab where you work as Research Engineer. Focus: industrial AI applications, vision-language models, academic publications. Complements your startup role with research rigor. Jul 15, 2026
  • › Substack Substack is a newsletter + blog publishing platform with built-in monetization, SEO, and distribution. Official 2024–2025 documentation specifies: 1,800–4,000 word ideal range for technical content; full markdown support including code blocks, callouts, tables, footnotes; 25 MB image limit with auto CDN; canonical URLs, OG/Twitter cards, auto sitemap; paid tiers with 0–100% free preview, referral program, per-post paywalls. Jul 15, 2026
  • › Target PhD Labs Curated shortlist of 20 labs for MS→Direct PhD applications (Fall 2026 start). Scored on research match, funding guarantee, advisor availability, placement record, and outreach responsiveness. Jul 15, 2026
  • › Ashish Vaswani Co-lead author of "Attention Is All You Need." Designed/implemented first Transformer models with Illia Polosukhin; crucially involved in every aspect. Co-founded Essential AI (enterprise LLMs). Jul 15, 2026
  • › X (Twitter) X is the real-time distribution channel for technical threads, paper announcements, and community engagement. 280 chars standard (25K for Premium Articles); threads get 2–3× engagement; media boosts impressions; optimal timing 7–9 AM / 5–7 PM PST weekdays for tech audience. Jul 15, 2026
  • › LangGraph A graph-based orchestration framework for building stateful, multi-step agentic workflows.