OxyJen

OxyJen is an open-source Java framework for orchestrating LLM workloads with graph-style execution, context-aware memory, and deterministic retry/fallback. You can treat LLMs as native nodes rather than simple helper utilities, allowing you to build multi-step AI pipelines that integrate cleanly with existing Java code. This framework provides you with the tools to manage complex AI workflows with a focus on reliability and developer productivity within the Java ecosystem. Most AI frameworks bolt LLMs onto existing utility patterns, a function call here, a helper class there. OxyJen takes the opposite approach: LLMs are native nodes. They live in the same graph as your data processors, validators, routers, and business logic. The framework doesn't care whether a node calls GPT-4 or reverses a string, it orchestrates all of them the same way. The result is pipelines that are debuggable, testable, and composable in a way that feels natural to Java developers rather than ported from Python. It supports sequential chains but now pipelines are also defined as graphs of NodePlugin instances connected by typed edges. Supports branching, routing, merging and parallel execution. Built in resilience features like jitter, retry cap, backoffs makes Oxyjen nodes deterministic and LLM output directly maps to Java classes/records with JSON Schema enforcements. Prompt is not hardcoded, PromptTemplate, Vairables and PromptRegistry helps in creating reusable prompts.