research / published
Applied machine learning for network quality, anomaly detection, LLM scoring, and efficient inference.
Parsa’s work connects objective telemetry, machine-learning models, and subjective user perception.
An open-source framework predicting QoE from delay, jitter, and packet loss while retaining up to 97% of ITU-T P.1203 prediction accuracy.
A survey and comparative analysis spanning simulation, measurement-driven, and hybrid anomaly-detection paradigms.
Operator-level aggregation using deterministic LLM comment analysis and network MOS comparison across approximately 48,000 live-stream comments.
A review of QoE frameworks and machine-learning algorithms for telecommunication networks.
Reproducible QoE assessment from network parameters, supported by automated data collection and controlled Docker network conditions.
A compact line-segment detector engineered for resource-constrained devices.
A survey of fuzzy-inference architectures, platforms, and emerging hardware trends.
Machine-learning-based quality-of-experience prediction in multimedia networks, affiliated with the Computer Networks Group.
Duke University · Credential focused on retrieval-augmented generation and Python.
Duke University · Credential focused on Rust and systems software development.