GitHubIndependent researchAI / ML
Mathematical misconceptions
Can a model recognize a misconception when the math question is new? Evaluation holds out entire questions and examines uncertainty and human review.
Software engineering / AI research
Raghu Ram
Thota.
I build the systems behind student assessment and study how AI can support the people who use them.
01Experience
From student testing
to results people use.
I work in Technical Outreach for Public Schools (TOPS), a program within the Center for Urban Affairs and Community Services (CUACS) at NC State University. I develop systems for testing, scoring, and reporting for K-12 students across North Carolina, serving the North Carolina Department of Public Instruction (NC DPI).
Senior software engineering & technical ownership
- Statewide systems
- From test administration to scoring and educator reports
- Full stack development
- Backend logic, databases, web interfaces, and integrations
- Technical ownership
- Client requirements, design decisions, mentoring, and support
Scoring, reporting, and everything that connects them.
I lead technical development across NC Check-Ins and summative assessment reporting. I build backend logic in PHP and Laravel, design database workflows, and develop interactive web interfaces with Vue.js, JavaScript, and jQuery. The work connects student records, scoring, educator reports, access controls, exports, and troubleshooting tools.
I helped move reporting from the legacy WinScan system to the web and added reports and capabilities that did not exist in the older system. I work directly with NC DPI staff to turn assessment requirements into application behavior and support the systems during testing periods.
Performance is one part of that work. Across Online Reporting and NCTest Admin, I improved SQL and batch processing so numerous overnight jobs went from hours to minutes.
PHP · Laravel · Vue.js · JavaScript · jQuery · SQL
- Assessment recordsStudent tests and responses
- Scoring and processingMatching, scoring, and reconciliation
- Review and releaseAuthorized checks for summative reports
- Educator reportsState, district, school, and class
Employer work is described at a high level. Student data and internal implementation details remain private.
02Research
Where can AI help?
And how will we know?
My engineering work led me to questions about scoring reliability, student understanding, and how models behave when the educational context changes.
Local models / assessment
AI assistance for written responses.
I initiated research into scoring constructed responses with local machine learning and language models. I examine agreement with human scores, partial-credit cases, serious errors, and uncertainty in both model predictions and historical scores.
Status: research and evaluation. AI does not determine official student scores.
Historical written responses
Human reference scores
Local ML and language models
Compare agreement and errors
When should a model's answer go back to a person?
Scoring reliability · Human review · Local evaluation
How I evaluate the work
Assumptions, evidence, and limits.03Public work
Open to examination.
Research code, methods, and selected application projects. Each repository shows a different part of how I work.
6 projects
Tutoring outcomesAI / ML
Trace the Ace
Methods for predicting tutoring outcomes when the learning context changes. Includes dialogue representations, calibration, fixed model blends, and synthetic examples.
Graduate researchResearch
ThesisEvosuite
Research into automated Java unit test generation and code coverage using EvoSuite.
Academic projectSoftware
Secure Banking System
A full stack banking application built with Spring Boot, Spring Security, Angular, and MySQL.
Application projectSoftware
Face Detection
A React and Express application that connects to the Clarifai API for face detection.
Earlier learning projectSoftware
Tic Tac Toe
A multiplayer game with real-time communication using Node.js, Express, and Socket.IO.
04About me
I like knowing how
the whole thing works.
My work has taken me from pretesting administration to scoring, reporting, and AI evaluation. I enjoy understanding how these parts connect, especially when a difficult issue crosses more than one system.
I grew up in Vijayawada, India, and came to the United States in 2017 for my master's degree. Today, I bring that same curiosity to software engineering and education research.
EducationM.S. Computer Science and Engineering
University of South Carolina · 2019
B.Tech. Computer Science and EngineeringJawaharlal Nehru Technological University · 2017
My technical toolkit
Application engineering
PHP, Laravel, REST APIs, Java
Data and reliability
SQL, MariaDB, MySQL, batch processing
Web interfaces
Vue.js, JavaScript, jQuery, Node.js
Research and evaluation
Python, FastAPI, local ML and language models
What I bring to a team
I talk directly with the people using the systems. Their requirements, difficult cases, and day-to-day questions shape the work.
05Let's connect
Good work starts
with a conversation.
Education software, a research question, or a technical challenge.
I would be glad to hear from you.