Data Engineer & Research Applicant

Hanumantha Rao Vulli

I build data systems and study how AI models behave when the data or evidence is imperfect.

I hold an MS in Data Engineering from the University of North Texas. My work brings together cloud data pipelines, software development and independent research in reliable AI, cybersecurity and scientific machine learning.

Based in Denton, Texas, USA · Open to research and engineering opportunities

Engineering meets inquiryCloud & data systemsCybersecurityScientific machine learning

What I’m researching

My independent studies compare existing methods, examine their mistakes and document what the results can support. Each project below links to its code, methods or written findings.

Phishing feature robustness

Independent study

How well does a phishing detector work when some of the information it relies on is missing? I compare models and study the errors caused by losing website or URL features.

Methods and limitations

Grouped historical-data tests and a paired screenshot/URL study examine feature reliability. In the paired study, URL ranking exceeded compact image and equal-fusion baselines. Dataset and collection limits remain; this does not establish performance on current attacks.

Read the study and evidence

Citation-support checkpoints

Independent study

When an AI answer includes citations, do those sources actually support it? I compare ways of flagging answers for review and examine simple factors, such as answer length, that can affect the results.

Methods and limitations

Uses ALCE sentence labels and 900 RAGTruth answers from 150 question groups. Length and generator controls complicate claims about the tested evidence-reading scores. The manuscript is a working draft, not peer reviewed.

Read the study and evidence

Early transient prioritisation

Independent study

Can the first few observations help identify which astronomical events deserve closer attention? I compare models that rank simulated events when follow-up resources are limited.

Methods and limitations

A RAPID-inspired GRU and forest model use matched three-day PLAsTiCC observations. In one independent simulated test, both missed the two targets at the 5% follow-up budget. No performance on observed astronomical events is claimed.

Read the study and evidence

These are independent studies using public datasets. Manuscripts are working drafts, not accepted publications. Results from historical or simulated data do not establish performance in real-world use.

Software and data projects

These projects turn questions about data quality, evidence and access into working local prototypes. The source repositories include instructions and recorded results.

Historical evaluation comparing three models by mean absolute error.
Chart from recorded evaluation · 84 historical test days
Data pipelines & evaluation

Mobility Quality Lab

A data pipeline and small analytics API built with historical bike-rental records. It checks incoming data and makes repeated loads safe.

View project and code
Methods and limitations

Uses 17,379 UCI Bike Sharing observations, transactional SQLite loading and 13 passing tests. A historical comparison covers 84 test days. The API runs locally; this is not a live forecasting service.

Read the evaluation
Workflow from Spark and Iceberg through dbt checks to operator approval.
Data quality workflow · recorded results in the demo
Data quality & approval workflows

Lakehouse Guardian

A workflow that checks data before publishing it to an analytics table, with an approval step for the person running it.

View project and code
Methods and limitations

Combines Spark, Iceberg, dbt, Airflow and LangGraph. Designed faults expose checks that work and errors they miss. Runs locally; AWS infrastructure was validated but not deployed.

Explore recorded results
Architecture showing host-assigned roles, allowlisted handlers and read-only storage.
Architecture summary · local read-only reference system
Controlled AI tool access

Catalog Sentinel

A controlled interface that lets an AI tool read approved data while blocking requests outside its assigned permissions.

View project and code
Methods and limitations

A read-only MCP server with role-based restrictions and an audit log. Small local tests check allowed and denied requests. This does not establish security for every model or provide remote authentication.

Workflow from hybrid retrieval through extractive citations to provenance review.
Citation workflow · recorded experiment in the demo
Retrieval & source provenance

Evidence Ledger RAG

A document-search tool that returns short extracts with exact source references, making the basis of an answer easier to inspect.

View project and code
Methods and limitations

Uses lexical search and source hashes. The evaluation contains 14 authored queries, and its support checks miss some changes in meaning. It returns extracts rather than evaluating freely generated answers.

Explore recorded results

DocForge AI

A document-generation prototype with streaming responses and checks for tracking usage and handling failed requests.

View project and code
Methods and limitations

A local application with SQLite accounting and optional test-mode checkout. Tests cover concurrent reservations, refunds and duplicate payment events. It is not a deployed commercial service.

Teaching and research writing

A practical SQL and Python lesson on trustworthy analytics, alongside an unpublished study of data contracts and evaluation leakage.

Read the teaching material
More to read

The teaching sample includes exercises, expected answers and a rubric. It is not a claim of teaching employment.

Read the study draft

Experience

Data Engineer · PRC Cloud Tech

USA · January 2026–present

AWS ingestion and ETL pipelines, reporting SQL, and troubleshooting services using Lambda, API Gateway and RDS.

Data Engineer · Tech Force

USA · September 2025–January 2026

Data pipelines, PostgreSQL and Oracle query tuning, data-quality investigation and delivery workflows.

Data Engineer · Infosys Now

USA · April–September 2025

Spring Boot services on AWS, data validation and workflow documentation.

Java Full Stack Developer · DBS Technologies

India · July 2022–May 2023

Angular interfaces, Spring Boot APIs, AWS deployment support and Jenkins/Docker automation.

Get in touch

For a research opportunity, an engineering role, or a conversation about the work above, email me.

vullihanumantharao@gmail.com