Mahmoud Khalil

Designed and implemented multimodal LLM pipelines for large-scale UI/UX evaluation of 1000 mobile app interfaces using Python.
Developed frameworks for prompt evaluation, zero-shot vs. few-shot benchmarking, and structured output parsing.

Designed and implemented multimodal LLM pipelines for large-scale UI/UX evaluation of 1000 mobile app interfaces using Python., Developed frameworks for prompt evaluation, zero-shot vs. few-shot benchmarking, and structured output parsing.

Built large-scale generative evaluation systems using 16 LLMs (including GPT-4, Gemini, Llama 3, Mistral) to assess content credibility in health videos.
Improved model response quality by 15% through refining 6 prompt engineering techniques.
Developed automated Python pipelines for data processing, transcription, and NLP analysis of 450+ health-related videos.

Built large-scale generative evaluation systems using 16 LLMs (including GPT-4, Gemini, Llama 3, Mistral) to assess content credibility in health videos., Improved model response quality by 15% through refining 6 prompt engineering techniques., Developed automated Python pipelines for data processing, transcription, and NLP analysis of 450+ health-related videos.

Identified at-risk students by analyzing academic data before final exams, enabling early academic support and intervention.
Tracked the multi-year growth of Elite-stream students across 11 schools to assess program success and analyze performance trends.
Built dashboards, and delivered insights and statistical summaries to school leaders and regional decision-makers.

Identified at-risk students by analyzing academic data before final exams, enabling early academic support and intervention., Tracked the multi-year growth of Elite-stream students across 11 schools to assess program success and analyze performance trends., Built dashboards, and delivered insights and statistical summaries to school leaders and regional decision-makers.

Developed, optimized, and evaluated ML models using TensorFlow for deep learning tasks (ANN, CNN), and scikit-learn for traditional ML algorithms (SVM, k-NN, Random Forest, and decision trees) for space debris classification.
Applied over-sampling for dataset balance, enhancing model precision by 74%.
Reduced feature dimensionality by 80%.
Assembled a real light curve time-series dataset containing 16,000+ entries of space objects, distinguishing it from simulated datasets.
Extracted 53 statistical features in Python and conducted exploratory data analysis (EDA) to gain insights into the dataset's characteristics.

Developed, optimized, and evaluated ML models using TensorFlow for deep learning tasks (ANN, CNN), and scikit-learn for traditional ML algorithms (SVM, k-NN, Random Forest, and decision trees) for space debris classification., Applied over-sampling for dataset balance, enhancing model precision by 74%., Reduced feature dimensionality by 80%., Assembled a real light curve time-series dataset containing 16,000+ entries of space objects, distinguishing it from simulated datasets., Extracted 53 statistical features in Python and conducted exploratory data analysis (EDA) to gain insights into the dataset's characteristics.