Quantum Research Collective
Pioneering hybrid
quantum architectures.
We bridge the gap between computational physics and advanced machine learning — building the tools that make quantum development accessible, reproducible, and impactful.
Our Origin
From Quantum Buddies to Ryoushi.
What began as a shared obsession between three researchers — scattered across Imperial College London, the University of Sheffield, and the University of Leeds — became something larger.
We started as Quantum Buddies, a name that captured our friendship. But our ambitions outgrew it. Ryoushi (量子, "Quantum" in Japanese) represents who we've become: a research collective that refuses to treat quantum computing as an abstraction.
We build things. We run experiments. We publish our failures alongside our successes. And we believe that the next decade of quantum computing won't be shaped by the biggest companies — it will be shaped by the sharpest communities.
The Problem
Three fractures in the quantum ecosystem.
Vendor Lock-In
Visual tools stay tied to one vendor's SDK. Learning is possible, but cross-provider building is exponentially harder. Every platform speaks a different dialect.
Fragmented Access
Different QPUs require different accounts, APIs, pricing models, and hardware-specific workflows. Running the same circuit on two backends shouldn't require two engineering efforts.
No Reproducibility
Experiments are hard to compare or build on because metadata, benchmarking, and result-sharing are inconsistent. Breakthroughs die as dead-end PDFs.
The Solution
One Workspace. Any Hardware.
Quaggle
The public hub to build, run, and rank quantum solutions. One workspace to design circuits, route jobs, capture reproducible runs, and publish benchmarkable results.
Design
Browser-native builder with parameters and broad format support.
Run
Submit to simulators or QPUs through one unified execution layer.
Verify
Capture seeds, logs, outputs, cost, and timing for every run.
Publish
Turn experiments into project pages with datasets and citations.
Compete
Launch leaderboards and verified profiles around real outcomes.
Product
Built for Reproducible Research
Simple enough for classrooms, strong enough for repeatable research workflows.
Core Builder
Supports OpenQASM, Qiskit, and Cirq with full parameter control.
Run Intelligence
Metadata-rich logs: backend, seed, shots, time, and cost per run.
Competition Engine
Leaderboards for accuracy, latency, and shot efficiency.
Publishing Layer
Project pages with dataset attachments and forking.
Talent Discovery
Verified profiles built on real quantum experiment history.
Enterprise Grade
Private workspaces, audit logs, and governed access.
Why Quaggle
Work faster. Get noticed.
- Easy setup & highly interpretable — Abstract vendor complexity across IBM, IonQ, Rigetti, QuEra, OQC and future providers without forcing users to abandon familiar SDKs.
- Route-to-best execution — Backend-aware transpilation and cost or latency estimation help users choose the cheapest or fastest viable path per workload.
- Reproducible by design — Pinned configs, immutable artifacts, seeds, logs, and content-addressed results make every run auditable and re-runnable.
- Connected with the research community — Public projects, datasets, competitions, and profiles create a content graph that gets more valuable as more builders participate.
Source: McKinsey Quantum Technology Monitor 2025, World Economic Forum Quantum Economy Blueprint
Active Research
Investigations in progress.
We are moving past theoretical whitepapers to engineer high-performance systems. Our open-source repositories focus on bridging classical machine learning architectures with quantum workflows, pushing simulation limits, and making quantum error correction scalable.
Syndrome-Net / QEC & Reinforcement Learning
Syndrome-Net isn't just another circuit simulator; it is a ground-up framework designed to fundamentally rethink how we approach quantum error correction. By treating decoding and calibration not as static lookup tables, but as continuous topological control problems, we open the door for adaptive, intelligent agents.
We have integrated state-of-the-art Deep Reinforcement Learning—specifically TITANS, Proximal Policy Optimization (PPO), and Soft Actor-Critic (SAC)—directly into the error correction loops. The framework fully supports surface codes, qLDPC, and advanced colour codes. It allows researchers to seamlessly iterate on new code discoveries, training autonomous models that can adapt to asymmetrical and fluctuating noise profiles inherent in modern NISQ hardware.
QuantumForge / HPC Simulation Engine
QuantumForge serves as the foundational High-Performance Computing (HPC) engine for our most complex algorithms, engineered to eliminate the bottleneck between high-level quantum abstraction and low-level classical execution. Written entirely in standard-compliant Rust, it acts as an ultra-fast backend capable of simulating 26-qubit statevectors directly on consumer-grade hardware.
By effectively circumventing the limitations of naive Python-based execution environments, QuantumForge delivers a nearly 4,000x speedup across crucial workflows. This bare-metal performance enables us to rapidly prototype and execute resource-intensive algorithms like the Generative Quantum Eigensolver (GQE) and Variational Quantum Eigensolver (VQE), driving precise ground-state estimations and complex molecular simulations.
Applied Machine Learning & Co-Design
Our ambition extends beyond theoretical physics into practical, high-impact utility. With a deep pedigree in classical artificial intelligence—spanning computer vision, physics-informed neural networks, and generative deep learning—we architect hybrid neural networks carefully tailored for high-dimensional and non-convex parameter spaces.
We are intensely focused on hardware-software co-design, studying how specific noise topologies dictate algorithmic feasibility. This approach translates directly to applications in combinatorial optimization and complex market simulations. By participating in global hackathons and open-source initiatives, we continuously bridge algorithmic blueprints with immediate, industry-facing challenges.
The Team
The Architects.
Sid Iliyasu
Imperial College London
PhD researcher in Robotics and MSc Design Engineering. Focused on practical use cases for quantum algorithms where product usability and real experimentation need to meet.
Dat Chi (Ryan) Le
University of Sheffield
Theoretical Physics, researching quantum machine learning at the Sheffield Quantum Centre. Co-founder of QNNOVATION — specialised quantum data for real-world problems. Exposed to both academic and industry-facing quantum work.
Gyanateet Dutta
University of Leeds
Electronics & Computer Science (Artificial Intelligence) with open-source, research, and hackathon experience spanning machine learning, quantum algorithms and computational software. Published work in computer vision, physics-informed neural networks, and quantum-hardware co-design.