Courses and events

A broad AI conference covering reasoning, planning, machine learning, NLP, vision and AI and society, with co-located applied and education programmes. Proceedings are open access on the AAAI site. Note that the site refuses automated checks, so this listing rests on a browser read.

The flagship natural language processing conference, with main, findings, demonstration and student research tracks plus workshops. Every paper is free to read in the ACL Anthology.

The largest computer vision conference, with more than 10,000 attendees, a big industry expo and a wide workshop programme. Papers are free to read on the CVF open access site.

Europe's leading computer vision conference, held in even years so that it alternates with ICCV. Proceedings are published by Springer with free copies on the ECVA site.

The second major NLP conference each year, with an emphasis on empirical and data-driven work. Main, findings and demonstration papers all appear free of charge in the ACL Anthology.

Google's annual developer conference, where Gemini, Gemma and AI Studio announcements usually land alongside Android and Web sessions. In-person attendance is by invitation, but the keynotes and every session are streamed free and remain on demand.

The biennial sibling of CVPR, run by the same community and alternating years with ECCV. The organiser's site showed two different date ranges for 2027 at the time of writing, 3-7 and 2-8 October, so check before booking.

A top-tier venue for deep learning and representation learning. Reviewing is open on OpenReview, so every submission and its reviews are public, and registration for 2027 opens on 23 December 2026.

One of the three main general machine learning venues, alongside NeurIPS and ICLR, with a main track, tutorials, workshops and an expo. Papers are published open access in the Proceedings of Machine Learning Research.

The main data mining and applied data science conference, with research and applied tracks, workshops, tutorials and the KDD Cup competition. Proceedings are in the ACM Digital Library.

A focused venue where machine learning meets computer systems, covering training and inference systems, compilers, distributed ML and hardware. Smaller than the big three, with open access proceedings.

The largest machine learning research conference, with a main track, a datasets and benchmarks track, tutorials, workshops and competitions. The 2025 edition in San Diego drew about 26,000 registrants and proceedings are free to read online.

NVIDIA's developer conference, with a CEO keynote, hundreds of technical sessions and hands-on training across agentic AI, AI infrastructure, robotics and CUDA. In-person passes are paid, while the keynote and virtual access were free in 2026 and sessions are available on demand afterwards.

The community conference for PyTorch and the open source AI stack around it, including vLLM, DeepSpeed and Ray. The 2026 edition reported more than 3,500 attendees, and all talks are posted free on the PyTorch YouTube channel afterwards.

A vendor-run practitioner conference for the Ray distributed compute ecosystem, with a training day followed by two conference days. The 2026 edition co-located the first vLLM Conference and covered foundation model training, multimodal data curation and LLM reinforcement learning.

Andrew Ng's three-course successor to the 2012 Machine Learning course, covering regression and classification, neural networks and decision trees, then unsupervised learning, recommenders and reinforcement learning. Each course can be audited free, with a paid subscription for graded work and certificates.

A catalogue of more than 100 short courses on topics such as prompting, agents, RAG and fine-tuning, many built with vendor partners, plus longer specialisations and professional certificates hosted on Coursera. Short courses are free to take, while Coursera certificates and a membership plan are paid.

A top-down course that starts by training working models and fills in the theory afterwards, built on PyTorch and the fastai library. Part 1 is the 2022 edition of nine lessons and Part 2 (2023) builds a Stable Diffusion pipeline from foundations. No newer edition had been published at the time of writing.

A dozen free courses built around the Hugging Face libraries, including the LLM Course, the Agents Course, the Deep RL Course, the Audio Course, the Diffusion Course and a post-training course. Content runs in Colab notebooks and is updated as the libraries change.

Eight video lectures that build backpropagation, an MLP language model, a transformer (GPT) and a BPE tokeniser from scratch in Python and PyTorch, with notebooks on GitHub under the MIT licence. The last lecture was added in February 2024 and the series is complete as published.

A one-week intensive of nine lectures and three software labs covering sequence models, computer vision, generative models, reinforcement learning and LLMs. All materials are open source under the MIT licence and the 2026 edition is online, with archives back to 2017.

Slides, notes and assignments for Stanford's NLP course are public, running from word vectors through transformers to large language models and a final project. Lecture videos from the 2024 offering are free on YouTube, and a paid professional version (XCS224N) exists through Stanford Online.

The lecture notes for Stanford's core machine learning course are public, covering supervised and unsupervised learning, learning theory and reinforcement learning. Full lecture videos from the 2018 offering are on Stanford Online's YouTube channel, while problem sets and recent recordings need a Stanford login.

NVIDIA's training arm, with a free self-paced catalogue alongside paid courses, instructor-led workshops and certification exams covering deep learning, generative AI, data science, accelerated computing and AI infrastructure. Workshops and exams are included with some GTC conference passes.

The public Discord through which the non-profit EleutherAI runs its open research on open models, interpretability, evaluation and open-weight safety. The server reported about 40,000 members in November 2026, and the institute also runs a summer mentorship programme for open research.

Hugging Face announces its hackathons and community events on its blog, which is the only stable listing. In 2026 these included the Build Small hackathon with Gradio, an agentic RL challenge with PyTorch and Unsloth, and the Open Together demo night that opened Open Source AI Week in October.

The main chat community for the Hugging Face Hub, libraries and courses, with channels for the LLM and Agents courses and for individual libraries. The invite reported about 248,000 members in November 2026. Use the huggingface.co join link, as the discord.gg vanity invite did not resolve.

Google's platform for machine learning competitions, datasets, hosted notebooks with a weekly GPU quota, discussion forums and short Learn courses. Since March 2026 anyone can host a community hackathon, and the Google and Kaggle five-day AI Agents intensive in June 2026 drew about 350,000 enrolments according to Google.

A practitioner community founded in 2020 around production machine learning, with a free Slack, a podcast, a newsletter, virtual tech talks and in-person meetups. In May 2026 it became the official user community of the Linux Foundation's Agentic AI Foundation.

Papers We Love is a volunteer network of reading groups that meet to present and discuss computer science papers, including machine learning, with talks dated through 2026 in a handful of active chapters. It is a good template for finding any local ML or PyData meetup, which usually run free on Meetup.com or Luma.