Weights & Biases builds an MLOps platform used to track experiments, version and iterate on datasets, evaluate model performance, reproduce models, visualize results and detect regressions. The platform emphasizes dataset interrogation to surface issues such as bias and personally identifiable information before production and was co-designed with partners and customers. It is integrated in over 20,000 open‑source repositories and cited in hundreds of academic papers, and is used by high‑profile model builders including OpenAI, Aleph Alpha, Cohere, Anthropic and Hugging Face. The company reports 700,000 users (up from 100,000 in 2021) and more than 1,000 paying users, and has a team of over 200 people based in San Francisco. Weights & Biases plans to launch Prompts, a product to monitor and evaluate LLM outputs and support fine‑tuning and prompt engineering workflows. Financially, the company has raised a total of $250 million to date and recently secured an additional $50 million tranche that values it at $1.25 billion. Weights & Biases (W&B) began as an experiment-tracking solution and has expanded across the machine learning pipeline to offer dataset versioning, model evaluation, data visualization, collaborative model training, reproducibility, and pipeline management. The platform serves more than 100,000 machine learning engineers and data scientists and is used by research and commercial organizations including OpenAI, Toyota Research Institute, NVIDIA, Facebook Research, Graphcore, and Pandora. W&B positions itself as a developer-first tool focused on making machine learning work in production, improving collaboration, iteration, and reproducibility. The company says it will continue building out its platform, add new functionality, and scale to meet the growing needs of the ML community. Founders Lukas Biewald, Shawn Lewis, and Chris Van Pelt are repeat entrepreneurs leading the company’s product and growth efforts. Weights & Biases provides an experiment-tracking platform and complementary tools for dataset versioning, model evaluation and pipeline management for machine learning practitioners. The company says it now has more than 70,000 users at over 200 enterprises and its user base grew roughly 60x in the last two years, according to an investor quote. Founded by Lukas Biewald, Chris Van Pelt and Shawn Lewis, the startup positions itself as focused on the ML practitioner rather than top-down enterprise sales. Customers include John Deere (robotics for targeted fertilizer application) and several pharmaceutical companies modeling molecular behavior. The company previously raised $20 million and will use the new funding to continue hiring across engineering, growth, sales and customer success. Weights & Biases describes itself as an early leader in the MLOps/experiment-tracking market. Weights & Biases provides development tools that let teams monitor and track machine learning model training by logging code, hyperparameters, and run metrics with minimal code changes. Its initial product captures training runs and related metadata to help reproduce and evaluate models. The company recently launched Benchmarks, a collaboration-focused product aimed at researchers to share and compare model results without rerunning experiments. Customers cited in the article include OpenAI, GitHub, Qualcomm, Toyota Research Institute, Stanford and Columbia. The founding team includes CrowdFlower founders Lukas Biewald and Chris van Pelt and former Google engineer Shawn Lewis; CrowdFlower (Figure Eight) was acquired by Appen for up to $300 million. Financially, the company completed a $15M second round and has raised $20M in total to date. Weights & Biases was founded by Lukas Biewald, Chris Van Pelt, and former Google engineer Shawn Lewis to build tools for machine learning developers. Its first product records and visualizes the process of training machine learning algorithms, letting developers revisit and share past experiments. The product is already being used by the nonprofit research company OpenAI. The company’s broader goal is to create a suite of development tools to streamline ML software development and make AI more accessible across industries. Weights & Biases aims to solve practical developer problems by working directly with practitioners on future products. Financially, the company has secured early institutional backing to fund product development.

Total raised

$215M

Funding rounds

5

Latest round

Equity

Latest activity

Aug 2023