Samah Alshrief, Ph.D.
AI data scientist, research-computing specialist, and higher-education policy analyst.
I help universities and researchers turn complex questions into working data and computing. I build research data services, private AI tools, and the infrastructure that makes computational research possible. A particular focus: connecting small institutions to the national computing resources their researchers already qualify for.
About
I'm a data scientist and research-computing specialist at Seton Hall University, where I help faculty, students, and whole institutions turn stalled research questions into working data pipelines, computing workflows, and evidence they can stand behind. I helped launch the university's first computational research projects, and I now co-lead its Research Computing Initiative.
My work sits where three fields meet. I hold a Ph.D. in Higher Education Leadership, Management, and Policy, with a concentration in research, assessment, and program evaluation, plus a Master of Science in Data Science and Engineering and a Master of Education. That mix is the point: I can build the technical thing and explain, in plain terms, why it matters to the people who fund it.
The through line is evaluation and evidence. I trained in quantitative methods at ICPSR's Summer Program at the University of Michigan, and my policy work runs from program evaluation and budget analysis to interrupted time-series studies of enrollment policy. Along the way I've contributed to more than $136,000 in funded projects, supported IRB applications for research teams, delivered over 100 data consultations, and coordinated 43 concurrent research projects. I serve as Seton Hall's ICPSR Official Representative and a Women in Data Science (WiDS) ambassador, and I mentor people moving into data careers.
Based in New Jersey and New York City. I work with teams on site and remotely.
Data management systems, research data services, and high-performance and cloud computing.
Program evaluation and policy research grounded in rigorous methods.
A practitioner inside the university, building research-computing capacity and training people to use it.
Document-grounded AI built on institutional data, with attention to privacy and ethics.
Work with me
Advising, training, and building for universities, research teams, government agencies, and organizations working with data and AI.
For small and mid-sized institutions without a research computing office: I connect your researchers to national resources like the OSPool and NSF ACCESS (much of it free to use), pick the right venue for each workload, and get the first pilots running so the capability outlasts my visit. I know both sides of this gap because I'm building the same capacity inside a university now.
Research data strategy, research-computing setup (HPC and cloud), data governance, and IRB-aligned data management. Helping teams stand up the infrastructure and practices that make computational research possible.
Hands-on sessions in data science and research tools (R, Python, STATA, ArcGIS, Atlas.ti, Qualtrics, research data management, and AI literacy), tailored to your team's projects and skill level.
Design and deployment of document-grounded, private AI tools over your institution's own data: retrieval-augmented systems that keep sensitive material in-house, as in my syllabus-analysis platform.
Talks and guest sessions on AI in higher education, research computing, private AI, and data ethics, for conferences, campuses, and organizations.
For agencies, institutions, and nonprofits: evaluation design, quantitative and qualitative analysis, and plain-language reporting on what a program actually does. My doctorate concentrated in research, assessment, and program evaluation, with ICPSR-trained methods behind it.
Methods and data support for doctoral students and research teams: study and survey design, quantitative and qualitative analysis, and getting messy data ready for a dissertation or a manuscript.
One-on-one mentoring for students and early-career researchers building data, computing, and AI skills, from a first analysis to a research-computing workflow they can run on their own.
See services and pricingRequest a consultationBring NARDC to your campus
Selected Work
A sample of applied projects across AI, research computing, and data science. The full record is in my CV.
Designed and deployed a private AI app for Business School faculty to analyze syllabi and generate document-grounded feedback, built with Python, Streamlit, ChromaDB, and a retrieval-augmented pipeline.
Part of the team forming a Research Computing Committee and an internal grant proposal for HPC clusters, high-performance storage, and scalable resources to support computational research across disciplines.
Helped launch the university's first cloud research-computing project (RStudio on AWS LightSail for Research) and taught sequence-data analysis using SRA and NCBI toolkits.
Built predictive models on 3D biomechanics data of baseball players, using regression, decision trees, random forests, and SVMs, to identify predictors of speed and reduce injury risk.
Coordinated 43 concurrent research projects, delivering data-management and analysis training across quantitative, qualitative, ArcGIS, and social-media methods, with one-on-one student consultations.
Served as data specialist for the School of Diplomacy, supporting Atlas.ti and Stata training and building lesson content with the policy librarian for international-relations research.
Background
Teaching
University courses and hands-on workshops in data science, research methods, and computing. Click a course to request its syllabus.
Writing & Talks
Plain-language articles and free tools for researchers, staff, and the people who run the systems behind them. Browse all writing or subscribe to follow along.
Which formats still open in twenty years, why Excel is a risky archive, and the two-copy rule for working and preservation files.
Read more →What ORCID iDs, DOIs, and ROR IDs are, why funders and repositories ask for them, and a short setup checklist.
Read more →What belongs in a dataset README, a plain-text template to copy, and the mistakes that make shared data hard to reuse.
Read more →What a data use agreement is, the clauses that cause trouble, who actually signs, why it takes weeks, and a checklist for getting one through your research office.
Read more →The federal three-year floor, the six-year misconduct window, IRB and FDA rules, the conflict with destruction promises in consent forms, and a five-question retention plan you can paste into a DMP.
Read more →What a consent form has to say for data sharing to be possible, the template phrases that quietly rule it out, sample wording to adapt with your IRB, and what to do with data you already collected.
Read more →What counts as a data management and sharing cost, where NIH and NSF want it in the budget, and sample justification language you can adapt.
Read more →What counts as an identifier, which rules apply (HIPAA, FERPA, the Common Rule), a step-by-step pass, and a quick test that tells you whether it worked.
Read more →What every variable entry needs, a template you can copy, how to draft one from your stats software, and the mistakes that make a codebook useless.
Read more →What findable, accessible, interoperable, and reusable ask of your data, why FAIR isn't the same as open, and a checklist for your next dataset.
Read more →The parts every data citation needs, APA and DataCite formats you can copy, and the mistakes that make a citation useless.
Read more →Faculty mini-grants, paid student Sprinternships, and a statewide course inventory across New Jersey's 18 community colleges, and where campus AI policy still has to fill in the gaps.
Read more →A $25 million statewide supercomputer deal with NVIDIA is coming. What's confirmed, what's still open, and how a small campus positions itself now.
Read more →The five statements that cover almost every paper, wording you can adapt, and the mistakes that get a manuscript sent back.
Read more →Pinning your software environment, setting random seeds, keeping raw data raw, and writing a README a reviewer can follow.
Read more →A practical guide for librarians and administrators at small campuses: start with DMP review, borrow national infrastructure, and grow one offering at a time.
Read more →The free federal dataset behind almost every college statistic: where to get it, and the definitions that mislead first-time users.
Read more →You have a cluster account and no idea what to type. Login nodes, schedulers, submit scripts, storage that isn't backed up, and the mistakes that kill first jobs.
Read more →Where the data go when the paper is done: discipline, institutional, and general repositories, the six things to check before you pick one, and how to share data about people without getting it wrong.
Read more →Not a ban and not a press release: what a real university AI policy covers, how it maps to policies you already have, and the one-page data grid that answers most questions.
Read more →Folder structure, file naming, versioning, a README that works, and backups that survive a lost laptop. Five decisions made once, at the start, that save the project later.
Read more →Your analysis won't fit your laptop and you have no campus cluster. How to tell whether the OSPool or NSF ACCESS fits your work, both free, and get an account this week.
Read more →Why "we ran the program and things got better" isn't an answer, the "compared to what" problem at the center of every evaluation, and which designs a real budget can actually support.
Read more →Your researchers likely qualify for national computing resources like the OSPool and NSF ACCESS at no cost. Why small campuses miss out, and how to get a first researcher running.
Read more →What consent actually covers, why dropping the name column doesn't de-identify anything, when secondary use needs another look, and what to check before data touches an AI tool.
Read more →The core skills worth teaching faculty and staff, a few durable questions to ask of any AI answer, and how to run a workshop that changes habits instead of just impressing the room.
Read more →What each required section of a DMP actually asks for, the mistakes that get plans flagged, and a reusable outline you can adapt per grant.
Read more →What private, document-grounded AI is, when a university should use it instead of a public chatbot, and how to build one responsibly.
Read more →Questions
A few things people usually want to know before reaching out.
Universities and colleges, research teams, government agencies, nonprofits, and doctoral students. The common thread is work that runs on data, research computing, or AI, whether that's setting up research data services, evaluating a program, or building a private AI tool.
Yes, that's a focus of my work. Many small and mid-sized campuses already qualify for national resources like the OSPool and NSF ACCESS, much of it free to use, but never get set up. I connect your researchers to the right resource, match each workload to the right venue, and get first pilots running so the capability stays after the project ends.
Private AI is a document-grounded system built over your institution's own material, so answers come from your data and sensitive content stays in-house. A public chatbot answers from general training data and sends your prompts to an outside service. When the material is confidential or governed by policy, a private, retrieval-augmented setup is usually the safer choice.
I'm based in New Jersey and New York City and work with teams remotely. Workshops and talks can be delivered on site or online depending on what suits your group.
Yes. Sessions in data tools, research data management, and AI literacy are tailored to your team's projects and skill level, including groups that are new to the material. The aim is habits people keep using, not a one-time demo.
Start with a short discovery call to scope your project or event. From there, engagements usually take the shape of a workshop or talk, a scoped project, or ongoing advisory. Use the contact form to tell me a little about what you need and I'll reply, usually within a couple of business days.
Contact
Tell me a little about your project or event and I'll reply, usually within a couple of business days.