Companies that use Weights and Biases

Source: Analysis of job postings that mention Weights and Biases (using the Bloomberry Jobs API)

My analysis shows that Weights and Biases purchases are driven primarily by engineering and data science leadership, with 60% of postings seeking Machine Learning Engineers who will use the platform daily. The remaining positions span backend engineers, technical program managers, and research scientists, all requiring MLOps expertise. Buyers prioritize infrastructure scalability, reproducibility, and streamlined model deployment across diverse use cases from autonomous vehicles to drug discovery.

Day-to-day users are hands-on ML practitioners building production pipelines. They leverage Weights and Biases for experiment tracking, dataset management, model versioning, and performance monitoring. I noticed many postings explicitly mention the tool alongside similar platforms like MLflow and DVC, indicating it's part of a standard MLOps stack. Users work with computer vision models, LLMs, and traditional ML across cloud environments, focusing on automating workflows from training through deployment.

The pain points reveal companies struggling to scale ML operations efficiently. One posting seeks someone to "streamline and automate the process of generating visualizations, dashboards, and reports using ML-based techniques." Another emphasizes "monitor model performance and manage models and datasets versioning to bolster operational efficiency." A third highlights the need to "set up infrastructure and practices for model tracking, versioning, and reproducibility." These companies need tools that bring order to complex ML lifecycles, ensuring models move reliably from experimentation to production while maintaining transparency and governance.

Source: Analysis of Linkedin bios of 7,388 companies that use Weights and Biases

I noticed that Weights and Biases customers are predominantly companies building AI and machine learning products as core infrastructure, not just using AI as a feature. These aren't traditional software companies dabbling in ML. They're deep tech firms developing autonomous vehicles (HoloMatic, Glydways), creating novel AI models and agents (Kunumi pursuing AGI, Simbol AI building visual reasoning models), building healthcare diagnostics (Freenome's blood tests, Hedera Dx's liquid biopsies), or providing MLOps and data platforms themselves (Innovaccer, Staircase AI). Many are in highly regulated or safety-critical domains like defense, healthcare, and financial services where model performance and reliability are non-negotiable.

The funding stages span a wide spectrum, from bootstrapped startups to post-IPO giants like Illumina and Dynatrace. However, the concentration sits firmly in Series A through C companies, those 50-200 employee sweet spots where ML infrastructure becomes critical but resources remain constrained. These are companies with enough traction to have real data problems (Kayrros raised $44M Series C, Innovaccer $275M Series F) but still moving fast enough that experiment tracking and model management create genuine competitive advantage.

Source: Analysis of tech stacks from 7,388 companies that use Weights and Biases

I analyzed companies using Weights and Biases and found they're overwhelmingly engineering-first organizations building ML/AI products. The combination of tools reveals teams that treat machine learning as core infrastructure, not a side project. These companies invest heavily in developer tooling and operational excellence, suggesting they're venture-backed startups with sophisticated technical teams building data-intensive products.

The pairing with Linear is particularly telling. Linear has become the project management tool of choice for engineering teams that value speed and simplicity over enterprise processes. Combined with Weights and Biases, it suggests ML engineers who want to track experiments with the same rigor they track code. The extremely high correlation with Docker Hub (66.4x) makes perfect sense too. ML workflows require containerization to ensure reproducibility, and teams doing serious model training need robust deployment pipelines. Retool's presence indicates these companies are building internal tools rapidly, likely dashboards for monitoring models or data pipelines without investing engineering time in custom UIs.

The full picture shows product-led companies in growth stage, probably Series A through C. The presence of Amplitude for product analytics and Sentry for error tracking reveals teams obsessing over product metrics and user experience. Pagerduty's inclusion, despite appearing in only 530 companies, shows the highest likelihood multiplier at 117.9x, which tells me these ML systems are mission-critical enough to warrant on-call rotations. This isn't academia or experimentation. These are companies where model performance directly impacts revenue.