Principal Data Scientist · Arizona State University
Hamidreza Ghasemi
I build machine learning systems end to end, from model development to production deployment, with a focus on deep learning, NLP and MLOps on the cloud.
I am Hamidreza Ghasemi (published as Hamidreza Ghasemi Damavandi), a Principal Data Scientist in the Actionable Analytics group within the Office of the University Provost at Arizona State University, where I oversee the complete lifecycle of machine learning projects, from model development to deployment. Previously, I was a Senior Machine Learning Scientist at Overstock.
I earned my B.S. in Electrical Engineering from the University of Tehran, ranked first overall in Iran, and my M.S. and Ph.D. in Electrical Engineering from the University of Iowa.
Research Interests
I like problems where rigorous methods meet real impact. These are the areas I keep coming back to:
Large Language Models & Generative AI
Retrieval-augmented generation, agentic workflows and LLM reasoning, and how to make them useful in real decision-making.
LLMsRAGAgentsReasoning
Trustworthy & Explainable AI
Uncertainty and confidence estimation, including step-wise confidence for LLM reasoning, plus model explainability with SHAP.
UncertaintyConfidenceSHAP
Statistical & Scientific ML
Statistical inference, applied statistics and scientific machine learning that respects the structure of the problem.
InferenceScientific MLApplied statistics
Deep Learning & NLP
Transformer architectures, representation learning and language understanding for text-heavy, real-world data.
TransformersNLPRepresentation learning
Signal & Image Processing
The foundation of my Ph.D. in Electrical Engineering: estimation, detection and computer vision.
EstimationComputer visionPh.D. research
Optimization & Reinforcement Learning
Constrained optimization (linear and integer programming) and learning to make sequential decisions under uncertainty.
Linear / integer programmingRL
Applied toStudent success & learning analytics at ASUAdvisor support with LLMsE-commerce pricing & discountsPublications on Google Scholar →
Experience
Oct 2022 — Present
Principal Data Scientist
Arizona State University · Office of the University Provost
Promoted from Senior Data Scientist. Leads the group's data science and ML work end to end, from model development to deployment.
Daily XGBoost persistence model with SHAP explanations, supporting proactive interventions for 100,000+ students
LLM solution with a RAG architecture on AWS (Bedrock, OpenSearch, Step Functions, Lambda) to help academic advisors reach students in need
End-to-end MLOps pipelines: deployment time from hours to minutes, 99% pipeline reliability
Automated monitoring with Slack alerts and Amazon Q, cutting incident response time by 50%
Multi-variate constrained optimization (linear & integer programming) to allocate discounts on sale items
Production Python pipelines on million-row datasets, containerized with Docker and orchestrated with Airflow
PythonOptimizationDockerAirflow
Mar 2018 — Mar 2020
Postdoctoral Research Associate
Arizona State University
Education
Ph.D., Electrical Engineering University of Iowa, Iowa City, IA
2016
M.S., Electrical Engineering University of Iowa, Iowa City, IA
2016
B.S., Electrical Engineering University of Tehran, Iran
2013
Systems & Projects
Systems I own
Arizona State University · Office of the ProvostIn production · daily
Student Persistence Prediction
A daily XGBoost model that predicts each student's likelihood of persisting, with SHAP explanations of the drivers behind low persistence, so advisors can step in early.
Scale100,000+ students reached through proactive interventions
ExplainabilityPer-student SHAP drivers alongside every risk score
OwnershipEnd to end: features, training, deployment, CI/CD and monitoring
OperationsDeployment cut from hours to minutes, 99% pipeline reliability, incident response 50% faster
Arizona State University · Office of the ProvostGenerative AI · AWS
RAG Assistant for Academic Advisors
An LLM solution built on a retrieval-augmented generation architecture that helps academic advisors proactively identify and assist students who need support.
UsersAcademic advisors working on student outreach
ArchitectureAmazon Bedrock for generation, OpenSearch for retrieval, orchestrated with Step Functions and Lambda
Production House-Price API with Missing-Data Imputation
Took a house-price model served by a REST API and made it production-ready: KNN imputation so estimates work with incomplete inputs, a redesigned request path that removed a serialization bottleneck, a batch endpoint, input validation, structured logging, Docker and CI.
Outside of work I love exploring the American West. A few favorite places:
Sedona, AZDevil's BridgeA hiker on Sedona's largest sandstone arch, red rock mesas behind.
Sand HarborLake TahoeSwimmers in clear turquoise water among granite boulders and pines.
Page, AZHorseshoe BendThe Colorado River wrapping around a sandstone butte, 1,000 ft below.
For Recruiters
Do you require visa sponsorship?
No. I am a U.S. citizen, so there is no visa, sponsorship or transfer involved. I earned an EB-2 National Interest Waiver green card on my own merit in 2016 and became a naturalized citizen in 2024.
Where are you based?
Tempe, Arizona (Phoenix area). I am happy to discuss remote, hybrid or on-site arrangements.
What kinds of roles interest you?
Senior and principal-level roles in machine learning, generative AI / LLMs and MLOps, especially where I can own models end to end, from problem framing and modeling to production deployment on AWS.
What are your strongest skills?
Building production ML systems: predictive modeling (XGBoost, deep learning), LLM and RAG applications, explainability with SHAP, and MLOps pipelines on AWS (SageMaker, Bedrock, Step Functions, Lambda).
Open to conversations about machine learning, MLOps and new opportunities. Send me a message here and it lands straight in my inbox, or pick a time that works for you.
