Summary
Saeed Khaki is a Senior Applied Scientist with 8+ years blending academia and industry to build production-ready NLP, CV, and ML systems, currently focusing on RLHF and vision-language alignment for Microsoft Copilot. He holds a PhD in Operations Research and Data Analytics and has led LLM finetuning, prompt engineering, and session-aware training to improve Alexa’s conversational shopping and intent understanding at Amazon. His toolset spans PyTorch, TensorFlow, Trax, Transformers, and statistical R packages, and he has hands-on experience deploying models on AWS and mobile via TensorFlow Lite. Notable projects include building Falcon-40b/Llama2-based chatbots, a hybrid BERT-MLP-Mixer for intent classification, and a semi-supervised image model for corn kernel counting that shipped in an internal app. Saeed pairs rigorous research (e.g., a custom 6-layer Transformer trained on CNN/DailyMail and a 95% NER LSTM) with practical production concerns like drift detection and bias reduction. Based in Iowa, he combines deep learning curiosity with a track record of moving experimental models into real-world applications.
8 years of coding experience
7 years of employment as a software developer
Doctor of Philosophy - PhD Operations Research and Data Analytics, Doctor of Philosophy - PhD Operations Research and Data Analytics at Iowa State University
Amirkabir University of Technology
Computer Vision Nanodegree Program Computer Vision, Computer Vision Nanodegree Program Computer Vision at Udacity
English, Persian