Building the database platform developers deserve.
We believe that modern applications shouldn't need separate databases for transactions, analytics, and AI. HeliosDB brings these capabilities together in a single, easy-to-use package that requires no operational expertise.
Modern applications have complex data needs spanning transactions, analytics, and AI.
┌─────────────┐ ┌─────────────┐ ┌─────────────┐
│ PostgreSQL │ │ ClickHouse │ │ Pinecone │
│ (OLTP) │ │ (OLAP) │ │ (Vectors) │
└─────────────┘ └─────────────┘ └─────────────┘
▲ ▲ ▲
└─────────────────┴─────────────────┘
Complex Sync
Multiple Teams
Higher Costs
Transaction Processing (OLTP) — User data, orders, sessions. ACID guarantees required. Low-latency reads and writes.
Analytics (OLAP) — Business intelligence, aggregations and reporting, time-series analysis.
AI & Vector Search — Semantic search, recommendation systems, RAG pipelines.
┌───────────────────────────────┐ │ HeliosDB │ │ OLTP + OLAP + Vector Search │ │ Single Database, Zero Config │ └───────────────────────────────┘
No complex sync between databases
Reduced operational overhead
Pay for one database, not three
We were building AI applications and found ourselves managing three or four different databases—PostgreSQL for user data, Redis for caching, Pinecone for vectors, and ClickHouse for analytics. Each had its own operational overhead, sync logic, and failure modes.
We asked: Why can't one database handle all of this?
So we built one.
Starting with a Rust-based storage engine on RocksDB, we added PostgreSQL wire protocol compatibility, native HNSW vector indexing, columnar storage for analytics, and unique features like database branching and time-travel queries. The result is a database that's as simple to deploy as SQLite but as powerful as a full data platform.
Zero configuration should be the default. Advanced options for those who need them, but batteries included for everyone else.
Developers shouldn't need to be database experts. Clear documentation, helpful error messages, and predictable behavior.
The core database is Apache 2.0. We believe in transparency and community. Enterprise features fund continued development.
Fast shouldn't mean fragile. We optimize aggressively but never at the cost of data integrity.
AI applications have unique needs—vector search, semantic caching, embedding storage. We build for this future.
Memory safety, performance, fearless concurrency
Battle-tested LSM-tree storage engine
Columnar format for analytics
State-of-the-art approximate nearest neighbor search
Compatibility with existing tools and ORMs
HeliosDB is open source under the Apache 2.0 license. Contributions are welcome.
You can audit every line of code storing your data
The project outlives any company
Better software through collaboration
Lower barrier to try and adopt
We're a team of database engineers, systems programmers, and developer advocates united by a passion for great developer tools.
Founder & CTO
Database Architecture & Distributed Systems
Distributed Systems
Community & Advocacy