Sahrish Mustafa

Agentic Dataset Generation: Designed and developed an agentic AI system to automate the construction of SWE-bench++, enabling large-scale generation and validation of C++ software engineering benchmark instances with minimal manual intervention.

Model Evaluation and Benchmarking: Evaluated and benchmarked multiple LLMs on C++ software engineering tasks, analyzing issue resolution accuracy, patch generation quality, and overall performance across the automated benchmark.

Automation Pipelines: Built scalable automation pipelines for repository analysis, environment provisioning, build execution, and evaluation, streamlining end-to-end dataset generation and benchmarking workflows.

Intern at SkyLabs AI (2025-06 – 2025-09)

• Agentic Dataset Generation: Designed and developed an agentic AI system to automate the construction of SWE-bench++, enabling large-scale generation and validation of C++ software engineering benchmark instances with minimal manual intervention.
• Model Evaluation and Benchmarking: Evaluated and benchmarked multiple LLMs on C++ software engineering tasks, analyzing issue resolution accuracy, patch generation quality, and overall performance across the automated benchmark.
• Automation Pipelines: Built scalable automation pipelines for repository analysis, environment provisioning, build execution, and evaluation, streamlining end-to-end dataset generation and benchmarking workflows.

Systematic Literature Review (SLR) Researcher at AI Security Research Collaboration (2025-06 – Present)

• HPC Security through AI Methods: Conducting research and writing a Systematic Literature Review (SLR) focused on the application of AI methods for High-Performance Computing (HPC) security, including threat detection, anomaly detection, and intelligent defense mechanisms.

Intern at Genesys Research Lab (2024-09 – 2025-12)

• Deep Learning Specialization: Completed the Deep Learning Specialization by DeepLearning.AI (Andrew Ng), covering neural networks, CNNs, sequence models, optimization, and deep learning best practices.
• Data Annotation and Dataset Creation: Performed data annotation, preprocessing, and dataset creation for computer vision and deep learning tasks to support model training and evaluation.

Bachelor of Computer Science in Computer Science – FAST National University of Computer and Emerging Sciences (2022-08 – 2026-06)