“hexDAG (soon to be) – new open source agentic framework”
HexDAG is a framework for building deterministic workflows with AI agents.
It combines the simplicity of rules with the flexibility of AI, imposing rigorous validation and controlling model stochasticity. Instead of chaotic agents or unpredictable multi-agents, HexDAG promotes declarative DAGs and async-first execution that automatically parallelize.
HexDAG’s philosophy is:
– Deterministic core (if-else > LLM)
– Declarative complexity (YAML is also code)
– Scalable infrastructure (retry, timeout, out-of-the-box cleanup)
The result? Agents that are fast, predictable, and testable, while LLMs are merely modules in a well-defined architecture.
Bio
I am passionate about AI and its potential to drive innovation.
As CTO I specialize in developing and deploying cutting-edge AI solutions at Omniviser driving better decision making. Our top-notch technology can help you cut costs, predict the future, and implement the innovation.
I hold a PhD in quantum gravity from the University of Warsaw, under the supervision of Krzysztof Meissner. I am deeply involved in theoretical physics research and have lectured at renowned institutions such as the Albert Einstein Institute (AEI) at Potsdam, Kyoto University, Massachusetts Institute of Technology (MIT), National University of Singapore (NUS), University of Cambridge, and University of Oxford.
As a Nvidia Ambassador, I collaborate with Nvidia to advocate for and showcase the exceptional capabilities of their GPUs in accelerating AI workloads. Through workshops, conferences, and webinars, I educate and inspire fellow AI enthusiasts and industry professionals to leverage data parallelism for optimal AI performance & implement generative AI solutions.