Khaled Mahmoud

Mercor.com: Improved AI models by providing feedback and evaluating agent correctness across diverse technical domains, including front-end development, back-end development, cloud computing, and software architecture.

Confidential Projects : Implemented AI guardrails and improved RAG accuracy by designing a chunking strategy that accounted for the structure of legal documents.

Responsibilities and Impact
Owned performance and platform engineering for Apache Spark and Generative AI workloads on Virtual and Bare-metal Clusters.
Built and configured Apache Spark + Hadoop clusters end-to-end (network, OS, and software), automating workloads with Python and Bash scripts, and integrating Spark metrics, CollectD, and Perfetto for performance analysis.
Automated execution and profiling of Generative AI workloads, collecting hardware performance counters via Intel emon and OS metrics, and developed a pipeline for end-to-end workload execution, data collection, and initial analysis.
Demonstrated 2× TPC-DS throughput by integrating Gluten + Velox into Apache Spark.