01
LLM-HyPZ
Published LLM-assisted pipeline for hardware vulnerability discovery.
2025 · research · Team
Python · LLMs · OpenAI embeddings · K-means · Prompt engineering
Abhi.
Data, AI, and bioinformatics for research that has to hold up.
scroll
01Selected work
Two systems worth inspecting.
Published pipeline and PersonaeAI synthetic survey research. Lab software follows.
02
PersonaeAI
LLM-powered synthetic survey research pipeline for human-vs-AI persona experiments.
2026 · research · Team
Python · Anthropic Claude · Pandas · ANES 2024 Pilot · Streamlit
Private
In the lab
2026
PDF QA System
Retrieval-augmented question answering over PDF corpora, including cybersecurity documents.
2026 · internal · Individual
Case study Private
2026
Systematic Review Processor
LLM-assisted literature platform scaled from a 20-paper pilot to 309 papers.
2026 · internal · Individual
2026
Sensitivity Analysis
R Shiny mixed-effects power and sensitivity app for hierarchical, clustered, and twin data.
2026 · internal · Individual
Private
2026
Associational Variable Analysis
R Shiny research software translating an associational methods framework into an interactive tool.
2026 · internal · Individual
Private
2026
Methylation Bioinformatics Pipeline
Lab bioinformatics pipeline for DNA methylation processing at LSCH.
2026 · internal · Team
Private
02About
A little about me.
Hey, I'm Abhi. I build research software and intelligent systems where statistics, AI, and real data have to work together.
Officially I'm a Scientific Analyst for Data, AI & Bioinformatics in the Laboratory for Social Connectedness & Health under David A. Sbarra. That means R Shiny research tools, LLM-assisted literature workflows, synthetic survey experiments, and computational pipelines for behavioral and health science — not slideware.
I earned my M.S. in Data Science at the University of Arizona with an NLP graduate certificate (GPA 4.0). Before that, a B.Tech in Artificial Intelligence and Data Science. I also write on Medium as @isjustabhi.
There's more to me than code. Badminton is how I unplug — the energy, the strategy, the competitive spark. Whether I'm debugging a model, shipping lab software, or on a court, I like learning, improving, and seeing what I can build next.
Data Scientist | AI | Bioinformatics
- Aug 2024 — May 2026
- M.S. in Data Science
- University of Arizona
- Graduate Certificate in NLP. GPA 4.0/4.0. Spring 2026 Distinguished Graduate Scholar, College of Information Science.
- Aug 2020 — May 2024
- B.Tech, Artificial Intelligence and Data Science
- Madanapalle Institute of Technology and Sciences (MITS), Andhra Pradesh, India
- Undergraduate degree in AI and Data Science. Affiliation listed on the IEEE IConSCEPT 2024 skin-disease classification paper.
03Writing
Notes from the build.
Actual posts from Medium. Nothing invented, nothing scraped from someone else.
04Research
Published work, with the citation you can copy.
LLM-HyPZ is the flagship paper. Bibliographic details come from IEEE Xplore, Crossref, and arXiv.
2025workshopFeatured
LLM-HyPZ: Hardware Vulnerability Discovery Using an LLM-Assisted Hybrid Platform for Zero-Shot Knowledge Extraction and Refinement
Yu-Zheng Lin, Sujan Ghimire, Abhiram Nandimandalam, Jonah Michael Camacho, Unnati Tripathi, Rony Macwan, Sicong Shao, Setareh Rafatirad, Rozhin Yasaei, Pratik Satam, Soheil Salehi
2025 IEEE International Conference on Data Mining Workshops (ICDMW), Washington, DC
LLM-HyPZ combines zero-shot LLM classification, contextualized embeddings, unsupervised clustering, and prompt-driven summarization to mine hardware-related CVEs at scale. On the 2021–2024 CVE corpus (114,836 entries), the framework identified 1,742 hardware-related vulnerabilities and distilled five recurring themes. LLaMA 3.3 70B reached 99.5% classification accuracy on a curated validation set. The pipeline surfaced 411 of 1,026 CVEs later used in MITRE MIHW 2025 analysis.
Contribution. Third author. Graduate Research Assistant in MITHRA Lab; IEEE lists College of Information Science, University of Arizona. Public sources describe contributions in data preprocessing, pipeline organization, RAG-related workflows, and experimental coordination. Exact task split is not itemized in the paper.
2024conference
Multi-Class Skin Disease Classification: A Study of Transfer Learning Strategies for Deep Learning Models
Paparao Mekala, B. Surendiran, P. Ramanathan, K. Ramanjaneyulu, S. Sai Manikanta Prakash, T. Reddy Venkata Chetan, N. Abhiram Varma
2024 International Conference on Signal Processing, Computation, Electronics, Power and Telecommunication (IConSCEPT)
The study uses a class-balanced 23-class dermatology dataset (192 training and 20 test images per class) with image processing and transfer learning. VGG19 and ResNet50 reach 83.69% and 82.82% accuracy respectively. This is a research prototype for classification, not a clinical diagnostic system.
Contribution. Co-author, listed as N. Abhiram Varma, Department of CSE (Data Science), Madanapalle Institute of Technology & Science. Author order places this as a collaborative undergraduate research paper.
2026mapping
Altering Social Connections to Improve Health? A Preregistered Systematic Mapping of Experimental Evidence
David A. Sbarra, A. Chowdhary, A. Farrell, E. Marshall, K. Nixon, S. Stanton, et al.
Deakin University / OSF (listed on Google Scholar, 2026)
Google Scholar lists this preregistered systematic mapping of experimental evidence on altering social connections to improve health. Full author order and Abhiram’s specific contribution should be confirmed against the OSF record before treating this as a peer-reviewed journal article.
Contribution. Appears on Abhiram’s Google Scholar profile. Author position and contribution statement need confirmation from the OSF project.
05Expertise
Capability, shown through evidence.
No percentage bars. Each item points to a paper, repository, or verified role from the latest resume.
Generative AI & NLP
Large language models
PersonaeAI (Claude); LLM-HyPZ; systematic review coder; private PDF QA
RAG & semantic retrieval
LLM-HyPZ RAG/embedding work; private PDF QA
Prompt engineering
LLM-HyPZ classification/summarization; systematic-review configurable prompts
Vector embeddings
OpenAI embeddings + clustering in LLM-HyPZ; resume lists vector embeddings / semantic retrieval
Classical NLP
TF-IDF on Housing student-care records; NLP graduate certificate
Statistics & Modeling
Mixed-effects models
LSCH power/sensitivity Shiny app with simr and lme4
Associational methods
AVA: semi-partial R², incremental validity, bootstrap indirect associations, 12-pattern framework
Power & sensitivity analysis
Simulation-based evaluation for hierarchical, clustered, and twin datasets
Hypothesis testing & regression
Listed on latest CV; used across LSCH and campus analytics roles
Programming & Data
Python
PersonaeAI, systematic-review-processor, Housing NLP, research pipelines
R
AVA and sensitivity Shiny apps; LSCH statistical workflows
SQL / Microsoft SQL Server
Housing & Campus Health ETL from SQL Server, SIS, and institutional platforms
ETL pipelines
Automated cleaning, QA, validation, and scheduled reporting for Housing OASI
Pandas / NumPy
Listed on latest CV; used in PersonaeAI and analytics workflows
Visualization & Apps
Tableau
10+ Housing dashboards; Campus Health; MeeBuddy rural KPIs
Power BI
Campus Health dashboards; Microsoft Power BI Data Analyst Associate
R Shiny
Private AVA and Sensitivity Analysis applications for LSCH
Streamlit / Plotly
PersonaeAI research dashboards for distributions and agreement
Full Stack
React + TypeScript
ReviewIQ, AI Student Care, AI Interview System
Next.js / Vite
Multiple deployed demos on Vercel
Python APIs
FastAPI systematic-review-processor; EarlyEco backend contributor
Research Engineering
Reproducible research software
AVA, sensitivity analysis, PersonaeAI, systematic-review-processor
Bioinformatics
Current LSCH title includes bioinformatics; DNA methylation pipeline maintained in role
Human-in-the-loop evaluation
PersonaeAI human-vs-synthetic experiments; LLM-HyPZ expert-guided prompting
Responsible data handling
Campus Health and Housing work under institutional privacy procedures
06Recognition
Awards with sources, not trophies.
2026
Distinguished Graduate Scholar, Spring 2026
University of Arizona · College of Information Science
Named on the College of Information Science Distinguished Graduate Scholars list for Spring 2026.
SourceSpring 2026
Co-Lab Award for Graduate Capstone
University of Arizona iShowcase · Applied Research & Industry Impact
Team award for a Spring 2026 graduate capstone in Applied Research & Industry Impact. Public evidence currently comes from LinkedIn and the GitHub profile; an official iShowcase winners page was not located.
Source07Experience
Official titles, then the work underneath.
Reconciled from the latest CV and resume. Self-reported metrics stay attributed.
Jun 2026 — Present
Tucson, AZ
Dr. David A. Sbarra, Professor of Psychology and PI
Scientific Analyst – Data, AI & Bioinformatics
University of Arizona · Department of Psychology · Laboratory for Social Connectedness & Health
Research staff building statistical software, AI/NLP workflows, and interactive analytics for behavioral science, psychology, and health research under David A. Sbarra.
- Lead technical development of Associational Variable Analysis (AVA), an R Shiny research tool for nonexperimental, cross-sectional data — zero-order and covariate-adjusted associations, multiple regression, semi-partial R², incremental validity, and bootstrap-based indirect-association analyses.
- Developed a 12-pattern classification framework across four associational conditions with automated interpretation, configurable thresholds, input validation, and interactive visualizations.
- Designed and deployed a mixed-effects power and sensitivity analysis application using R, simr, and lme4 for hierarchical, clustered, and twin datasets.
- Built PersonaeAI, an LLM-powered synthetic survey research pipeline with Python, Anthropic Claude, Pandas, and the ANES 2024 Pilot dataset for 1,500 respondent profiles.
- Ran human-vs-synthetic survey experiments reporting a 33% reduction in mean total variation distance with politically enriched AI personas versus baseline.
- Independently developed an LLM-assisted systematic literature review platform and expanded literature processing from a 20-paper pilot to 309 research papers.
- Collaborate with academic researchers and quantitative methodologists on reproducibility, validation, and interpretability.
20 → 309
Literature scale-up
Self-reported on the latest CV/resume
−33%
Synthetic survey TVD
Self-reported on the latest CV/resume (PersonaeAI)
Title and bullets reconciled from the latest CV/resume. LinkedIn may also list Data Scientist @ University of Arizona | LSCH.
Jun 2026 — Present
United States
Founder
Stealth Startup
Independent startup focused on technology-driven solutions. Product details stay private while the company is in stealth.
- Founded an independent startup focused on developing technology-driven solutions.
Aug 2025 — May 2026
Tucson, AZ
Linda Scheu, Director of Organizational Assessment & Strategic Initiatives
Graduate Assistant – Data Science
University of Arizona · Housing & Residential Life (OASI)
Graduate assistant supporting institutional assessment, housing operations, and student-support analytics with dashboards, ETL, and NLP on care records.
- Developed and maintained 10+ interactive Tableau dashboards for institutional assessment, student engagement, housing operations, and strategic planning, with automated daily refreshes.
- Designed automated reporting workflows that reduced manual reporting effort by approximately 60%.
- Built NLP workflows with Python and TF-IDF to analyze unstructured student-care records and summarize themes related to stress, mental health, and physical well-being.
- Extracted, cleaned, transformed, and integrated datasets from Microsoft SQL Server, SIS, and institutional platforms into scheduled ETL pipelines.
- Collaborated with Housing & Residential Life, Campus Health, and the Dean of Students on dashboards and reports under responsible data-handling practices.
10+
Dashboards
Latest CV/resume
−60%
Reporting effort
Self-reported on the latest CV/resume
May 2025 — Aug 2025
Tucson, AZ
Rachel Abraham, MPH, Evaluation Manager
Data Analyst
University of Arizona · Campus Health Services
Healthcare data analyst supporting evaluation of student health programs through analytical datasets, dashboards, and reports.
- Designed interactive Tableau and Power BI dashboards covering utilization, mental health, AOD, program participation, and student health trends.
- Contributed to survey reporting improvements associated with a reported 25% increase in survey response rates.
- Ran cohort comparisons, EDA, and trend assessments in Python and SQL to evaluate health-program outcomes.
- Integrated assessment, survey, and health-information sources with cleaning, consistency checks, and validation under institutional privacy procedures.
+25%
Survey engagement
Self-reported on the latest CV/resume
Aug 2024 — Aug 2025
Tucson, AZ
Dr. Rozhin Yasaei, Assistant Professor and Director, MITHRA Lab
Graduate Research Assistant
University of Arizona · MITHRA Lab · College of Information Science
Graduate research assistant on LLM-assisted hardware vulnerability discovery, contributing to LLM-HyPZ (IEEE ICDMW 2025) and related LLM4Sec workshop work.
- Contributed to LLM-HyPZ, applying LLMs and information retrieval to hardware cybersecurity vulnerability discovery over 114,836 CVE records (2021–2024).
- Supported large-scale preprocessing, dataset organization, experimental coordination, RAG/embedding experimentation, and prompt-driven classification.
- Supported research contributing evidence to the MITRE 2025 Most Important Hardware Weaknesses initiative.
- Co-authored the IEEE ICDMW 2025 paper and contributed to research presented at the ICDM 2025 LLM4Sec Workshop.
- Collaborated with interdisciplinary researchers, including contributors affiliated with Intel, on project development and documentation.
Jan 2023 — Jul 2024
Remote · India
Data Analyst
MeeBuddy · Rural India's Digital Buddy
Data analyst supporting engagement analytics and rural KPI dashboards for a digital companion product.
- Analyzed user engagement, service adoption, and platform utilization across rural segments with Python, SQL, and EDA.
- Built interactive Tableau dashboards tracking activity, engagement trends, and digital-service adoption.
- Translated segment insights into recommendations supporting digital-service access in rural and underserved communities.
08Beyond the code
Sometimes I put the laptop away.
When I'm not debugging models, you'll probably find me on a badminton court.
Badminton is how I unplug. I like the pace of it — the small decisions, the footwork, the way a rally asks you to stay present after a day of staring at a model.
I play for the energy and the competitive spark, not a ranking. No titles to list. Just a court, a shuttle, and a reason to close the laptop.
Photos from actual games can live here later. Until then, the court lines are the picture.
Move across the court. Tap the lines a few times if you want the easter egg.
09Open source
Public repositories, with a static fallback.
Fetched from the GitHub API when available. No token is used in the frontend.
10 · Contact
Collaborations, roles, and a conversation.
Open to AI engineer, applied scientist, ML engineer, research engineer, and data scientist roles where systems meet real data and real users.
abhiramn@arizona.edu+1 (484) 447-2590
Ask about Abhiram
Grounded answers only.
This assistant searches a curated public record: titles, papers, repos, and contact details. It does not call an external model and will not invent credentials.
Abhiram Varma Nandimandalam (Abhi) is a Scientific Analyst for Data, AI & Bioinformatics in the Laboratory for Social Connectedness & Health at the University of Arizona Department of Psychology, under David A. Sbarra. He builds research software spanning AVA, mixed-model sensitivity analysis, PersonaeAI synthetic surveys, and LLM-assisted literature review, and co-authored LLM-HyPZ (IEEE ICDMW 2025).