Draft:Harshil Patel
Submission declined on 13 April 2026 by VortexPhantom (talk).
Where to get help
How to improve a draft
You can also browse Wikipedia:Featured articles and Wikipedia:Good articles to find examples of Wikipedia's best writing on topics similar to your proposed article. Improving your odds of a speedy review To improve your odds of a faster review, tag your draft with relevant WikiProject tags using the button below. This will let reviewers know a new draft has been submitted in their area of interest. For instance, if you wrote about a female astronomer, you would want to add the Biography, Astronomy, and Women scientists tags. Editor resources
|  |
Comment: In accordance with Wikipedia's Conflict of interest guideline, I disclose that I have a conflict of interest regarding the subject of this article. KUNAL PAI (talk) 22:11, 12 April 2026 (UTC)
 | This article has multiple issues. Please help improve it or discuss these issues on the talk page. (Learn how and when to remove these messages)
|
Harshil Patel is a computer scientist and software developer known for his contributions to computer architecture simulation and multi-agent artificial intelligence systems. He is a key contributor to the open-source gem5 simulator and a co-creator of several AI research frameworks, including NAAMSE and HASHIRU.
Education
[edit]Patel attended the University of California, Davis, where he earned a Bachelor of Science in Computer Science in 2023. He is currently completing a Master of Science in Computer Science at the same institution, maintaining a 4.0 GPA while serving as a Graduate Student Researcher.[1]
Career and Research
[edit]Computer Architecture Simulation
[edit]From 2023 to 2025, Patel worked as a software developer at UC Davis, where he made significant contributions to gem5, a widely used open-source computer architecture simulator heavily utilized by academia and industry leaders.[1] His role as a full-time contributor to the project was officially recognized by the ACM Special Interest Group on Computer Architecture (SIGARCH) in 2024.[2] His work focused on automating workflows for building disk images and benchmark suites, as well as migrating the gem5 resources database to Microsoft Azure to improve scalability.
He co-developed the gem5 Vision Project alongside Parth Shah, Kunal Pai, and Arslan Ali, which created the official resources.gem5.org web portal.[3] This project improved user-friendliness and accessibility by introducing advanced search functionality, comprehensive resource categorization, and expanded database support within the gem5 ecosystem for researchers and developers. The system's source code was subsequently open-sourced,[4] and the work was presented at the gem5 Workshop at the International Symposium on Computer Architecture (ISCA) in 2023.[5]
Patel subsequently co-authored research on standardizing reproducible simulation workflows across ISAs, which was accepted to the IEEE International Symposium on Performance Analysis of Systems and Software (ISPASS) in 2026.[6]
Artificial Intelligence and Multi-Agent Systems
[edit]Patel's recent research focuses on the intersection of AI, multi-agent systems, and security evaluation, with a strong emphasis on open-source reproducibility.
HASHIRU: In early 2025, Patel co-developed HASHIRU (Hierarchical Agent System for Hybrid Intelligent Resource Utilization), a budget-aware multi-agent orchestration framework.[7][8] The system dynamically decomposes tasks and routes them to specialized local and cloud-based AI agents based on capability and cost, minimizing resource overhead for complex reasoning tasks. The entire framework was released as an open-source project to facilitate further research in hybrid agent orchestration,[9] and has been highlighted by AI research discovery platforms such as Emergent Mind as a key model for dynamic agent lifecycle management,[10] tool creation,[11] and multi-agent assistant systems.[12]
NAAMSE: Later in 2025, he co-created the Neural Adversarial Agent Mutation-based Security Evaluator (NAAMSE).[13] Patel engineered the behavioral scoring engine for this autonomous red-teaming tool, which uses evolutionary algorithms to generate adversarial prompts against large language models.[14][15] The project won second place in the Agent Safety Track at the UC Berkeley RDI AgentBeats competition in February 2026.[16] The framework's novel use of fitness-guided search over mutations for uncovering AI vulnerabilities has been featured in third-party analyses of advanced threat frameworks for autonomous agents by platforms such as Emergent Mind.[17] The framework was subsequently accepted to the Agents in the Wild workshop at the International Conference on Learning Representations (ICLR) in 2026.[18]
Selected Publications
[edit]- Reproducibility and Standardization in gem5 Resources v25.0 (ISPASS 2026)
- NAAMSE: Framework for Evolutionary Security Evaluation of Agents (ICLR 2026 Workshop on Agents in the Wild)
- HASHIRU: Hierarchical Agent System for Hybrid Intelligent Resource Utilization (arXiv, 2025)
- gem5 Vision (ISCA 2023: gem5 Workshop)
References
[edit]- 1 2 "Harshil Patel". LinkedIn. Retrieved 2026-04-12.
- ↑ "What's New in gem5". ACM SIGARCH. 2024-07-26. Retrieved 2026-04-12.
- ↑ "gem5 Resources". gem5.org. Retrieved 2026-04-12.
- ↑ "gem5/gem5-resources-website". GitHub. Retrieved 2026-04-12.
- ↑ Shah, Parth; Pai, Kunal; Patel, Harshil; Ali, Arslan (2023). "gem5 Vision" (PDF). gem5.org. Retrieved 2026-04-12.
- ↑ Pai, Kunal; Patel, Harshil; Le, Erin; Krim, Noah; Samani, Mahyar; Bruce, Bobby R.; Lowe-Power, Jason (2026). "Toward Reproducible and Standardized Computer Architecture Simulation with gem5". arXiv:2512.13479v2 [cs.AR].
- ↑ Pai, Kunal; Shah, Parth; Patel, Harshil (2025). "HASHIRU: Hierarchical Agent System for Hybrid Intelligent Resource Utilization". arXiv:2506.04255 [cs.AI].
- ↑ "HASHIRU: Hierarchical Agent System for Hybrid Intelligent Resource Utilization". Emergent Mind. Retrieved 2026-04-12.
- ↑ "HASHIRU-AI/HASHIRU". GitHub. Retrieved 2026-04-12.
- ↑ "Dynamic Agent Lifecycle Management". Emergent Mind. Retrieved 2026-04-12.
- ↑ "Tool Creation Agent". Emergent Mind. Retrieved 2026-04-12.
- ↑ "Multi-Agent Assistant System". Emergent Mind. Retrieved 2026-04-12.
- ↑ "NAAMSE: Framework for Evolutionary Security Evaluation of Agents". hashiru-ai.github.io. Retrieved 2026-04-12.
- ↑ Pai, Kunal; Shah, Parth; Patel, Harshil (2026). "NAAMSE: Framework for Evolutionary Security Evaluation of Agents". arXiv:2602.07391 [cs.CR].
- ↑ "NAAMSE: Evolving Security Tests for AI Agents". Emergent Mind. Retrieved 2026-04-12.
- ↑ "Agentic AI Weekly: Berkeley RDI February". Berkeley RDI Substack. 18 February 2026. Retrieved 2026-04-12.
- ↑ "Advanced Threat Framework for Autonomous AI Agents (ATFAA)". Emergent Mind. Retrieved 2026-04-12.
- ↑ Pai, Kunal; Shah, Parth; Patel, Harshil (March 2026). "NAAMSE: Framework for Evolutionary Security Evaluation of Agents". OpenReview. arXiv:2602.07391. Retrieved 2026-04-12.
- provide significant coverage: discuss the person in detail, not brief mentions or interviews lacking independent analysis;
- are reliable: from reputable outlets with editorial oversight;
- are independent: not connected to the person, such as interviews, press releases, the subject's own website, or sponsored content.
Please add references that meet all three of these criteria. If none exist, the subject is not yet suitable for Wikipedia.