The team

Built by people who ran bandits in production

BanditDB is a spinoff from Dynamic Pricing AI. Two co-creators, several years of contextual bandits in live systems, and the one database we wished we had from the start.

Co-creators

Two people, one loop.

Co-creator

Simeon Lukov

Co-creator of BanditDB. Built and ran contextual bandit models for live pricing decisions at Dynamic Pricing AI before turning that experience into a database.

Co-creator

Georgi Angelov

Co-creator of BanditDB. Developed and operated bandit models in production at Dynamic Pricing AI, and co-designed the engine that replaced the patchwork of tools behind them.

Origin

From a patchwork of tools to one database.

  1. 01 · Dynamic Pricing AI Years in production

    At Dynamic Pricing AI the team developed and ran bandit models in production for several years, making pricing decisions on live traffic.

  2. 02 · The friction Fragmented tools

    The model lived in one place, its state in another, outcome logging and offline evaluation somewhere else. Every new decision meant wiring the same pieces together again.

  3. 03 · The product One decision database

    So they sat down and built BanditDB: predict, act and reward in one engine, with model state, durability and evaluation built in.

Say hello

Talk to the people who built it.

Questions about bandits, campaigns or running BanditDB in production. The co-creators answer on Discord.