Human–AI Interaction
Investigating how interface design, evidence presentation and explanations influence comprehension, verification behaviour and appropriate reliance on generative AI.
OPEN TO MRES & PHD RESEARCH OPPORTUNITIES
HELLO, I'M AMRUTHA VARSHINI REDDY THALLA
Building AI that people can understand and trust.
My research direction is Human–AI Interaction: how people interpret, verify and make decisions with AI-generated information. I bring an MS in Information Systems, practical GenAI projects and software engineering leadership to this question.
HUMAN + AI
01 / EVIDENCEWhere does this answer come from?
02 / UNDERSTANDINGCan a person verify it?
SourcesClarityTrust
RESEARCH IN PROGRESS / 2026
PRIMARY FIELDHuman–AI InteractionExplainable AITrustworthy GenAIHuman-centred evaluation
01 / RESEARCH DIRECTIONMaking intelligent systems
Making intelligent systems
meaningful to people.
I am interested not only in what AI systems produce, but in what their users understand, question and ultimately decide.
Explainability & visual communication
How can citations, visual explanations and uncertainty cues make AI outputs easier to evaluate?
Trust calibration
When do people over-rely on AI, and which designs help them recognise limitations?
Evidence-grounded GenAI
How should RAG systems expose their supporting sources and communicate missing evidence?
High-impact applications
Exploring these questions in health information and other contexts where understanding matters.
PROPOSED RESEARCH APPROACH
Design → Evaluate → Learn
Literature review, interface prototyping, comparative studies, comprehension tasks, source-verification measures and careful analysis of user trust. These are proposed methods for future supervised research, not claims of completed studies.
02 / SELECTED WORKFrom building systems
From building systems
to asking better questions.
A selection of hands-on technical projects and academic work that informs my research direction.
EVIDENCELENS / PROTOTYPE
How can we assess an AI answer?
Evidence source 01
Explanation claritySource checking
RESEARCH-ORIENTED PROTOTYPE01
EvidenceLens
An exploratory interface for evidence-aware question answering, explanation comparison and demonstration of human evaluation workflows. It does not represent a completed participant study.
Human–AI InteractionRAGStreamlit
RETRIEVE → GROUND → EXPLAIN
GENERATIVE AI02
Research-paper Q&A with RAG
Built a question-answering pipeline over research papers using embeddings, ChromaDB retrieval and source-linked responses to support checking answers.
PythonChromaDBLangChain
BASE MODEL
↓
LOW-RANK ADAPTATION
↓
STRUCTURED OUTPUT
MODEL ADAPTATION03
LoRA fine-tuning
Fine-tuned a compact language model for structured answers and compared its outputs with an unchanged baseline, including examples outside training.
TransformersPEFTEvaluation
DATA → MODELS → INTERPRETATION
APPLIED MACHINE LEARNING04
Healthcare insurance analytics
Analysed insurance records, compared regression models and developed dashboards and an interactive interface to communicate estimated charges.
Scikit-learnXGBoostPower BI
IS 7000 · MASTER'S CAPSTONEGift card management
Gift card recordView ↗
Gift card recordView ↗
ACADEMIC CAPSTONE05
Collaborative web application
Contributed a React gift-card listing interface with Axios API calls, state management, conditional rendering and error handling in a team-based capstone.
ReactREST APIsGit / Jira
View documented pull request ↗AGENT A
⇄
AGENT B
COMPARATIVE EXPLORATION
AI FRAMEWORKS06
Multi-agent framework comparison
Implemented comparable research-and-summary workflows in CrewAI and phidata, observing how search-tool differences affected output quality.
CrewAIphidataStreamlit
03 / BACKGROUNDGrounded in engineering.
Grounded in engineering.
Moving toward research.
My professional experience informs how I approach usable, reliable and well-explained technology.
Education
2023 — 2025
MS, Information Systems
Indiana Institute of Technology, USA
GPA 3.7 / 4.0 · Master's capstone in collaborative web developmentCompleted 2018
B.Tech, Computer Science & Engineering
JNTUH, India
Academic projects: GPS route navigation; online examination systemCONTINUING TECHNICAL PREPARATION
Full Stack Data Science & Generative AI training, Naresh i Technologies · December 2025 – June 2026
Professional experience
DEC 2025 — PRESENT
Team Lead
IPlex Technologies Private Limited
Team and product-development coordination across QuickBuddy and QuickBooking, including software workflows, delivery planning and cross-functional collaboration.FEB — NOV 2025
Software Engineer
Cortracker Inc
SQL Server, ASP.NET Core, REST APIs, reporting dashboards and software delivery for HR and CRM applications.TECHNICAL TOOLKIT
Python · SQL · C# / .NET · React · RAG · LangChain · Streamlit · Git · Machine Learning
04 / CONNECTInterested in research
Interested in research
that puts people first?
I welcome conversations with prospective supervisors and research groups working in Human–AI Interaction, HCI, explainability and responsible generative AI.