PostgresML
PostgresML is an open-source Postgres extension that integrates data storage and machine learning inference directly within the database. It leverages GPUs for accelerated computations, enabling in-database ML/AI operations, including support for Large Language Models (LLMs) from Hugging Face and a Retrieval-Augmented Generation (RAG) pipeline with functions for chunking, embedding, ranking, and transforming text. The platform offers over 47 classification and regression algorithms, boasting 8-40X faster inference compared to HTTP-based model serving and supporting millions of transactions per second with horizontal scalability. It also provides NLP tasks like text classification, translation, and summarization. PostgresML aims to simplify AI application development by eliminating the need for separate systems and data transfers, enhancing performance, security, and scalability.
PostgresML
PostgresML is an open-source Postgres extension that integrates data storage and machine learning inference directly within the database. It leverages GPUs for accelerated computations, enabling in-database ML/AI operations, including support for Large Language Models (LLMs) from Hugging Face and a Retrieval-Augmented Generation (RAG) pipeline with functions for chunking, embedding, ranking, and transforming text. The platform offers over 47 classification and regression algorithms, boasting 8-40X faster inference compared to HTTP-based model serving and supporting millions of transactions per second with horizontal scalability. It also provides NLP tasks like text classification, translation, and summarization. PostgresML aims to simplify AI application development by eliminating the need for separate systems and data transfers, enhancing performance, security, and scalability.
TL;DR
Product: Open-source Postgres extension for in-database ML/AI with GPU acceleration and RAG primitives
Key capabilities: In-database LLM inference (Hugging Face), vector search integration, chunk/embed/rank/transform pipeline
Performance: Claims of 8–40x faster inference vs HTTP model serving and support for high TPS with horizontal scalability
Funding: Seed round (May 2023), ~$4.7M total reported
Company Overview
Problem Domain
In-database machine learning and LLM inference; retrieval-augmented generation and vector search for NLP workloads.
Founded
2022
Industry
Machine learning / Database infrastructure
Tech Stack
Funding Track Record
4700000.00
Investor roster reported to include several angels and early-stage investors alongside Amplify Partners
Investor Signal
“Seed led by Amplify Partners with participation from angel/individual investors (e.g., Jack Altman, Max Mullen, Brandon Leonardo, Rafael Corrales, Jeremy Stanley, James Yu, Greg Rosen)”