Microsoft · April 2026 – Present
Principal ML Engineer & Technical Leader
LLMs · NLP · Speech/ASR · AI Evaluation · Responsible AI
Microsoft Atlassian Amazon/AWS
I build and lead production AI systems from modeling and evaluation through deployment and scale.
Across Microsoft, Atlassian, and Amazon/AWS, my work has spanned conversational AI and ranking, Responsible and Generative AI, developer AI, and now speech recognition—building systems that move from research and evaluation into products operating at scale.
I stay hands-on across modeling, evaluation, architecture, and production engineering, while providing technical direction across complex AI initiatives. My background combines Computer Engineering with a Master’s in Computational Linguistics from the University of Washington.
Progression across production AI systems, technical leadership, and hands-on ML engineering.
April 2026 – Present
Speech / AI Systems
Working on production speech recognition and AI systems, spanning quality and evaluation, reliability, latency, scalability, and production engineering.
May 2025 – April 2026
Developer AI
Led developer-AI initiatives across Rovo Dev, including Code Search and AI-assisted software development, and contributed to Rovo AutoReview / Code Reviewer for automated PR review.
June 2022 – May 2025
GitHub Copilot · OpenAI / GPT-4 Evaluation · Bing Chat / Copilot
Part of the team that built GitHub Copilot, leading Responsible AI capabilities spanning harmful-content safety, privacy/PII, malware, vulnerability detection, and protected material/IP safeguards. Led cross-organization work across GitHub, Bing, Azure OpenAI, and OpenAI, including evaluation and benchmarking of models such as GPT-4 for Bing Chat/Copilot.
2017 – 2022
AWS AI · Amazon Lex · Amazon Ads
Built production NLP capabilities for Amazon Lex, including natural-language data augmentation that expanded small sample-utterance sets with diverse linguistic variation to improve intent-model robustness and accuracy. Designed and launched a semantic-relevance ranking feature across Amazon Ads auctions, partnering with economists and Platform, Targeting, and Auction teams; the launch drove approximately $750M in additional revenue.
2014 – 2017
Windows feedback intelligence and Azure Spark / HDInsight platform work.
2013 – 2014
Web-scale multilingual NLP for Office and knowledge graph systems.
2011 – 2013
Co-founded a healthtech startup building clinical workflow software and mobile applications; led a small engineering team.
Selected production AI systems spanning safety, developer AI, conversational AI, and ranking.
Microsoft
Part of the team that built GitHub Copilot, leading Responsible AI and security capabilities across harmful-content safety, PII/privacy, malware, and software-vulnerability detection. Drove cross-organization execution across GitHub, Bing, Azure OpenAI, and OpenAI, and led evaluation and benchmarking of OpenAI models including GPT-4 for Bing Chat/Copilot.
Microsoft
Led technical work behind Protected Material detection supporting Microsoft’s Customer Copyright Commitment for generative AI. Built safeguards for generated text and code using lexical and semantic detection, working across Bing, GitHub, Azure OpenAI, and OpenAI.
Atlassian
Led Code Search work for Rovo Dev’s developer-AI stack and contributed to Rovo AutoReview / Code Reviewer for automated PR review. Atlassian later reported a year-long Code Reviewer evaluation across 1,900+ repositories, with a 30.8% reduction in median PR cycle time and 35.6% fewer human-written review comments.
AWS AI · Amazon Lex
Built natural-language data augmentation for Amazon Lex, expanding small customer utterance sets with generated linguistic variation—including typos, contractions, morphological variants, and emojis—to create richer training data and improve Bi-LSTM intent-model robustness and accuracy.
Amazon Ads
Designed and launched a semantic-relevance penalty for Amazon Ads auctions that reduced exposure of ads poorly matched to customer queries and product context. Partnered with economists and Platform, Targeting, and Auction teams across the Ads stack; the launch drove approximately $750M in additional revenue.
Public research and patents connected to secure, edit-time code AI.
Patent
Vulnerability reduction for syntactically incomplete codeGranted patent covering vulnerability reduction in syntactically incomplete code during edit-time developer workflows.
Paper
Transformer-based Vulnerability Detection in Code at EditTime: Zero-shot, Few-shot, or Fine-tuning?Co-authored work on detecting software vulnerabilities directly on syntactically incomplete code at edit time, comparing zero-shot, few-shot, and fine-tuned transformer approaches for real-time developer workflows.
University of Washington, Seattle
Graduate work spanning NLP, model evaluation, and responsible and inclusive language technologies.
Cairo University · Faculty of Engineering
Coursework completed; thesis not completed.
Cairo University · Faculty of Engineering
Graduated with honors; ranked 4th in class.
Selected panels, talks, and technical discussions.
Panel
Reimagining the Future of Software Delivery with AIPodcast
Responsible AIArabic-language technical discussion on Responsible AI and practical considerations around building and deploying responsible AI systems.
Domains where I have deep production experience.
Available for select advisory and consulting engagements where deep AI/ML expertise can help shape technical strategy, evaluation, and production systems.
Technical direction, system architecture, build-vs-buy decisions, and moving AI products from prototype to production.
Evaluation strategy for LLMs, speech, and production AI systems—from metric design and benchmarking to launch-quality assessment.
Practical approaches to AI risk, safety evaluation, safeguards, and production readiness.
Technical guidance across language models, NLP, speech/ASR, developer AI, and related production systems.
For advisory engagements, speaking, professional opportunities, or other inquiries, send a note here or reach me on LinkedIn.