PostgresML
PostgresML is an open-source extension for PostgreSQL that enables machine learning directly within the database. It allows users to train, deploy, and run machine learning models using SQL queries, eliminating the need to move data between the database and external ML tools. This integration helps reduce latency, improve performance, and simplify the architecture for ML-driven applications.
We’re a humble team of engineers, academics and creatives primarily located on earth.
As a Founding Product Designer I shaped the product’s design strategy and created a seamless user experience that integrated machine learning into everyday workflows. I built a scalable design system and worked closely with Product Managers and Engineers to make sure our teams were aligned. I also set up processes to improve how we worked together, streamlining collaboration between engineering and product.
Key features
In-Database Machine Learning: Train and serve models without leaving PostgreSQL.
Support for Popular Libraries: Integrates with ML libraries like scikit-learn, XGBoost, and more.
Real-Time Predictions: Enables real-time inference through simple SQL queries.
Scalable Architecture: Optimized for high performance with support for large datasets.
Our story
PostgresML was founded in 2022 by the engineers who built (and rebuilt) Instacart’s machine learning platform.
In the midst of the Covid-19 pandemic — and the company’s highest-ever growth period — they found that the data infrastructure for machine learning at scale was too complex and data hungry to reliably deliver fast and efficient systems.
They created a platform that eliminated microservices and moved the models right into the database. They saw dramatic performance improvements to their search and recommendation apps along with an increase (from 9 months to 1 week!) in development cycles.
PostgresML was created to provide the same critical infrastructure to today’s ML/AI apps and their practitioners. Whether it’s RAG or search; PostgresML users can access the same in-database platform to set up, train, test and run state-of-the-art models in a single solution.
Case studies
-
![]()
Onboarding
We aimed to simplify onboarding but faced challenges that hindered user adoption. Issues like mandatory credit card entry, an ineffective marketing survey, and unclear post-signup guidance created friction. Through targeted improvements, we revamped the onboarding experience, resulting in increased signups, better engagement, and a more seamless user journey.
-
![]()
Multi-org + Teams
Multi-org focuses on implementing an industry-standard capability that supports the creation of multiple projects with distinct team members, thereby streamlining the team creation process. However, this opportunity introduces several complexities that must be addressed to ensure a seamless user experience and maintain system integrity.
-
![]()
Billing
The Billing feature provides users with comprehensive control over their organization’s financial management. Users can easily add and manage payment methods, settle past due notices, and gain clear visibility into their expenses. This level of transparency empowers users to monitor usage, optimize spending, and maintain financial accountability across their organization.
-
![]()
Notebooks
Notebooks is an interactive workspace that allow users to combine code, queries, visualizations, and text in one place. Designed to make data exploration and analysis more seamless by enabling users to manipulate data, and visualize results directly within the notebook. Making this a powerful tool for data-driven teams who need to analyze, document, and share their work efficiently.
Design system
