Open-source frameworks, peer-reviewed methods, and end-to-end systems, built by the
current cohort and used by researchers beyond LDQIS.
The software behind the lab's studies: frameworks, testbeds, and experiments with research questions and papers behind them.
Federated Learning, Agentic AI
Knowledge Management Framework for Federated Learning
A configuration and execution framework for federated learning simulations built on Flower. Implements novel PID controller-inspired and Trust-based aggregation strategies alongside classical robust methods (Krum, Multi-Krum, Bulyan, RFA, Trimmed-Mean), adversarial attack scheduling, federated LLM fine-tuning with LoRA, and a wide range of medical imaging datasets. Published in IEEE Intelligent Systems, 2026.
Built with Python, Flower, PyTorch, LoRA
2,471 experiments run so far with InteFLAgentic AI, Software Engineering
A pre-registered study of what coding agents learn from code review
A coding agent becomes useful once it knows how an organization writes software: its internal APIs, error-handling patterns, naming rules, and what its reviewers reject. Sphragis tests whether that house style can be learned at all from an organization's public code-review history, and what a declared boundary protects once it can. The study is pre-registered: its pass rule is fixed in code and its test window is sealed before collection. It is written as a Registered Report for MSR 2027.
Built with Python, PyTorch, Transformers, PEFT
Federated Learning, Intelligent Security
A labeled fleet testbed for federated personalization
Public corpora lack what a disclosure boundary needs: objects that carry real classification levels and cross-cutting compartments. Pharos generates a labeled body of reporting in which two holders can be incomparable rather than merely ranked, so the personal/shared split in federated personalization can be measured rather than asserted. A static explorer lets anyone browse the corpus.
Built with Python, PyTorch, PEFT, OpenTelemetry
Federated Learning, Adversarial Robustness
Full-stack robustness platform, forked from InteFL
A platform for stress-testing federated learning under attack: configure, run, and compare experiments across 9 aggregation strategies, 11 adversarial attack types, and 20+ datasets, all from a React + FastAPI + Celery web app with live, streamed run monitoring. InteFL remains the lab's canonical research framework; Phalanx-FL is its robustness and tooling proving ground.
Built with Flower, FastAPI, React, Celery
Agentic AI, Sensemaking
Sensemaking for nonatomic entities using AI coding agents
Evidence about a single entity is scattered across graph, relational, and text stores, and analysts lose the thread at every pivot between systems. Ariadne uses an agentic harness built on the Claude Agent SDK as one analytic interface over that infrastructure: a single agent traces an entity's evidence across heterogeneous stores into a cited analytic narrative.
Built with Python, Claude Agent SDK, Neo4j, MCP
Federated Learning, Quantum ML
When do variational quantum circuits keep federated training data private?
Federated clients share gradients instead of data, and on classical networks those gradients can be inverted to rebuild the training inputs. For variational quantum circuits the literature disagrees. This project maps where each claim holds, across encoding depth, trainable layers, batch size, and simulated IBM Heron device noise, using gradient-matching attacks on 4- and 8-qubit circuits. A course project for CSCI-739 Quantum Machine Learning, Fall 2026.
Built with Python, PennyLane, Qiskit, JAX
Federated Learning, High-Performance
Rust-core federated learning orchestration
A federated learning runtime with a Rust aggregation core and a Python-first interface. Eight paper-cited aggregation algorithms (FedAvg, FedProx, Krum, Bulyan, Geometric Median, and more), round-level and data-pipeline attack simulation, and aggregation kernels measured at ~92× the speed of pure Python.
Built with Rust, PyO3, Python
Agentic AI, Multi-Agent System
A multi-agent system with three interfaces
Ten specialized A2A-protocol agents (planner, coder, tester, reviewer, scribe, validators) collaborate in a shared group chat to plan, build, and ship code, with every step observable through OpenTelemetry. One backend drives three interfaces: a terminal CLI, a Pygame GUI, and a Ren'Py visual novel. Presented at North-East AI Agents Day 2026.
Built with MCP + A2A, CLI, Pygame, Ren'Py
Open-source tools for working with AI coding agents such as Claude Code, built for the lab's own research and free for anyone to use.
Agentic AI, Developer Tools
A Claude Code plugin marketplace
Plugins for Claude Code: skills for code repositories (toolchain audits, CI log review, adversarial pre-merge review, cross-repo drift) and for documents (papers, slides, bibliography verification), opt-in git guards, and a spoken Claude Code.
Built with Python, Claude Code
Agentic AI, Developer Tools
Gives Claude Code a voice
After each Claude Code reply, keryx says its gist out loud in one or two sentences: what happened, and what Claude needs from you. A local model writes the summary and a local voice speaks it, so it costs no tokens.
Built with Python, Ollama, Kokoro
Developer Tools
Your Claude Code usage limit, always on screen
A thin always-on-top bar for Windows and Linux showing how much of your Claude Code usage limit you have used, fed by the status line Claude Code already renders. No API calls and no credentials.
Built with Rust, Tauri
