Independent Student-Led Research Group · University of Maryland Global Campus
Aura Computing & Intelligence Lab
Keeping pace with rapid advancements in AI and data technology. We explore practical and reliable intelligence systems — from large language models and intelligent agents to software engineering — tackling critical technical and real-world challenges through open experimentation.
From our work
Aura Lab Technical Report · 2026
Benchmarking Agentic Tool Use and Reasoning in Repository-Level Code Comprehension
Aura Lab Preprints · 2025
Structured Retrieval and Factuality Verification for Scientific Domain Large Language Models
Open Source Software & Datasets · 2026
Aura-Bench: An Open Evaluation Suite for Multi-Step AI Agent Tool Execution
Research areas
AI & Large Language Models
Reasoning, retrieval-augmented generation, evaluation, and domain-specific applications.
Intelligent Agents
Autonomous planning, tool use, structured workflows, and digital environments.
AI for Software Engineering
Code understanding, debugging, testing, architecture review, and mobile systems.
Human–AI Interaction
Usability, reliability, transparency, and human-agent collaborative systems.
Recent work
-
Aura-Bench: An Open Evaluation Suite for Multi-Step AI Agent Tool Execution
Open Source Software & Datasets · 2026
-
Benchmarking Agentic Tool Use and Reasoning in Repository-Level Code Comprehension
Aura Lab Technical Report · 2026
-
Structured Retrieval and Factuality Verification for Scientific Domain Large Language Models
Aura Lab Preprints · 2025
What we build
Our research translates algorithmic ideas into verifiable, open experimental tooling and applied systems.
Research Prototypes
Working systems that translate conceptual models into testable, measurable, and reproducible experiments.
Open Tooling & Benchmarks
Publicly shared evaluation harnesses, datasets, and automation suites supporting peer replication.
Applied AI Systems
Multi-agent workflows and developer tooling tailored for complex codebase understanding and real-world software tasks.
Recent news & milestones
Aura-Bench v0.2 Published
Released open evaluation suite for multi-step AI agent tool execution and sandboxed trace analysis.
Structured Retrieval & Verification
Preprint on factuality verification for scientific domain LLMs made available.
Multi-Agent Code Intelligence Project
Launched experimental testbed for autonomous regression testing and patch localization.
Life in the lab
Learning data and AI skills allows us to gain deeper insights and tackle critical challenges across digital technologies and intelligent automation. Beyond theoretical research, Aura Lab fosters collaborative experimentation, practical toolkits, and learning by making.
Explore lab life