Marqov offers an agent‑driven orchestration platform that automates hybrid quantum‑classical workloads by containerizing each computational step and recording complete provenance metadata for reproducible execution. The system integrates with leading quantum simulators and HPC schedulers, providing a REST API and web dashboard for workflow management, versioning, and collaborative sharing.
Funding
Funding not disclosed
Founders
Product
Problem
Computational research, especially projects that combine quantum simulators with classical high‑performance computing (HPC), suffers from fragmented toolchains and undocumented environment changes, leading to results that cannot be reliably reproduced. The lack of systematic orchestration and provenance capture makes it difficult for researchers to validate findings or share workflows across institutions.
Solution
Marqov delivers an agent‑driven orchestration platform that automates the execution, monitoring, and versioning of complex quantum‑classical hybrid workloads. By encapsulating each computational step in reproducible containers and recording full execution metadata, the system creates an immutable provenance graph that can be replayed on demand. Integrated adapters connect to leading quantum simulators, cloud‑based HPC schedulers, and on‑premise clusters, enabling seamless hybrid job scheduling from a single interface. The platform’s API and web dashboard expose real‑time status, resource utilization, and deterministic checkpointing, allowing researchers to verify and iterate on experiments without manual reconfiguration. Built‑in compliance with FAIR data principles ensures that results are discoverable, accessible, and reusable across collaborative projects.
Target Audience
The primary customers are computational scientists, quantum algorithm developers, and research institutions that run hybrid quantum‑classical experiments and require rigorous reproducibility guarantees. Additionally, HPC centers and cloud service providers offering quantum workloads benefit from Marqov’s orchestration and provenance capabilities.
Features
- Agentic workflow engine that dynamically allocates quantum and classical tasks based on resource availability and workload dependencies
- Containerized execution environment with automatic capture of software stack, hardware configuration, and random seed values for full reproducibility
- Immutable provenance ledger that logs input parameters, intermediate data artifacts, and execution timestamps, exportable in JSON‑LD or RDF formats
- Native connectors for IBM Q, Rigetti, AWS Braket, as well as Slurm, PBS, and Kubernetes HPC schedulers
- RESTful API and Python SDK for programmatic workflow definition, version control, and integration into CI/CD pipelines
- Web‑based dashboard offering visual workflow graphs, real‑time performance metrics, and one‑click rollback to prior experiment states
- Role‑based access control and end‑to‑end encryption to protect sensitive research data while supporting collaborative sharing