Python NLP React LLM FastAPI RAG Whisper Groq pgvector
NLP RESEARCHER — HYDERABAD, INDIA
Yamini G

NLP Researcher focused on Sentiment Analysis & Intent Modeling

SYSTEMS  ·  ALGORITHMS  ·  NLP  ·  INTELLIGENT AI

I study how AI systems fail when surface-level sentiment masks deeper user intent. My research sits at the intersection of NLP, intent modelling, and LLM reliability — three preprints published on TechRxiv (IEEE) and SSRN, seven deployed AI systems built from research to production.

◎  Hyderabad, India

Researcher. Builder.
System Thinker.

I am a CS graduate specialising in NLP research, LLM reliability, and the design of AI systems that go beyond surface-level language understanding.

My research investigates a specific failure mode — when AI systems respond to how something is said rather than what it means. Three preprints that formalise this gap and propose intent modeling frameworks that operate beyond sentiment detection.

Alongside the research, I build production systems that operationalise these findings — clinical NLP pipelines, hallucination benchmarking platforms, RAG systems with claim-level verification, and interview-coaching and journaling platforms driven by intent-modeling pipelines.

7
Deployed AI Systems
3
Research Papers
IEEE · SSRN
Published
Curiosity

What My Research Is Actually About

Most NLP systems are built to detect sentiment. What they cannot do reliably is infer intent — the difference between someone saying "I'm fine" and meaning it, and saying "I'm fine" while meaning something else entirely. My research formalises this gap, identifies its failure modes, and proposes frameworks that treat intent as a first-class signal rather than a downstream assumption from sentiment.

CURRENT FOCUS
Applied NLP & LLM Systems

Studying failure modes in sentiment-aware systems, benchmarking LLM hallucination across model families, and building production-grade NLP pipelines grounded in published research frameworks.

Skills & Stack

LANGUAGES
Python JavaScriptTypeScript
AI & NLP
NLPLLM Integration Prompt EngineeringRAG Groq APISentiment & Intent Modeling Speech-to-Text
FRONTEND
React.jsNext.js HTML5CSS3 Tailwind CSSFramer Motion
BACKEND & DB
FastAPIDeno REST APIPostgreSQL / pgvector MongoDBRedisSupabase
CLOUD & DEPLOY
VercelRender DockerGitHub Actions GitGitHub

Theory, Tested.

01 / RESEARCH-BACKED

MindNook

Sentiment-Aware Reflective Writing System

Deployed implementation of my TechRxiv (IEEE) framework — a five-layer pipeline operationalizing sentiment detection, pragmatic classification, temporal trend recognition, goal alignment, and utility-based response selection in a live journaling platform.

NLPSentiment Analysis LLMSupabaseFull-Stack
  • Implements all 5 layers of published framework
  • Utility-theoretic intervention threshold (τ*)
  • Individual sentiment baseline (z-score) per user
  • Pragmatic speech-act classification
  • Configurable AI sensitivity (C_fp / C_fn)
02 / EVALUATION PLATFORM

LLM Reliability Lab

Medical QA Hallucination Benchmarking

Browser-based platform benchmarking LLM reliability on medical question answering, classifying hallucination types, and comparing prompting strategies across model families. Chain-of-thought prompting roughly halved the hallucination rate versus zero-shot on this benchmark.

Next.jsGroq API LLaMA 3MixtralGemma 2
  • Cross-model, cross-prompt comparison
  • 3-category hallucination taxonomy
  • Live inference via Groq Cloud API
  • Research write-up included (PDF + LaTeX)
  • 20-question curated medical QA benchmark
03 / RESEARCH INTELLIGENCE

Prism

RAG Platform with Claim-Level Verification

Full RAG pipeline — hybrid dense (pgvector) + BM25 retrieval fused via reciprocal rank fusion, cross-encoder reranking, streamed grounded generation, and claim-level hallucination detection — with the retrieval layer exposed: chunk similarity scores, source attribution, and per-claim grounding status.

Next.jsFastAPI pgvectorGroq API
  • Claim-level hallucination detection
  • Exposed retrieval layer (similarity scores)
  • Structured summarization (TLDR/methods/results)
  • Hybrid dense + BM25 retrieval with RRF fusion
  • Multi-format ingestion (PDF/DOCX/URL/text)
04 / INTENT MODELING

InterviewIQ

Interview Simulator with a Four-Layer Intent Model

Interview practice and simulation platform where every spoken or typed answer is passed through a four-layer Multi-Layer Intent Model (MLIM) — affective sentiment, pragmatic speech-act classification, goal-state tracking, and intent fusion — alongside conventional question generation, scoring, and a browser-side integrity-monitoring layer.

Next.jsFastAPI MongoDBGroq API
  • Four-layer MLIM pipeline on every answer
  • Practice and Simulation interview modes
  • Voice answers via Web Speech API + Whisper transcription
  • Browser-side integrity monitoring (tab/DevTools/inactivity)
  • JWT auth with refresh-token rotation and reuse detection

4 of 7 deployed systems shown above — explore the rest on GitHub.

View All Projects on GitHub ↗️

Research Work.

TechRxiv · IEEE Preprint · 2026

A System-Level Framework for Sentiment-Aware Reflective Writing Systems: Modeling Temporal Emotional Patterns with Interpretability and Ethical Safety

This paper analyses the limitations of sentiment-based AI systems and proposes a multi-layer framework integrating pragmatic reasoning, temporal pattern recognition, and goal-aware intent interpretation for reflective writing contexts.

Read Preprint
TechRxiv · IEEE Preprint · 2026

Beyond Surface Affect: Why Sentiment Detection Alone is Insufficient for Intent Interpretation in Human–AI Communication

This research formally distinguishes sentiment detection from contextual intent interpretation in human-AI communication systems, proposing a richer framework for understanding user intent beyond surface-level sentiment signals.

Read Preprint
SSRN · Preprint · 2026

Comparative Sentiment Analysis of YouTube Transcripts and User Comments: Failure Modes and Interpretability in Public Discourse

This research presents a dual-model sentiment analysis system combining LSTM neural networks with VADER lexicon-based analysis to compare creator transcript sentiment against audience comment sentiment across YouTube content domains, revealing systematic divergence patterns and documenting domain-specific failure modes.

Read Preprint

Blog & Articles.

PHILOSOPHY · ENGINEERING
Why Every Engineer Should Study Philosophy Before Building Anything
On how philosophical frameworks — Kantian ethics, virtue ethics, Socratic questioning — can help engineers examine assumptions, ethical consequences, and system objectives, and ask whether something should be built, not only whether it can be.
Read Article ↗
PRAGMATICS · AI
If AI Can't Handle "I'm Fine," It Can't Handle the Future
On the gap between semantic and pragmatic understanding in conversational AI — indirect meaning, conversational context, hedging, uncertainty, and emotional subtext — and why pragmatic reasoning matters for reliable AI systems.
Read Article ↗
COMING SOON
The Sycophancy Problem: Why Your Chatbot Agrees With Everything You Say
In progress…
Coming Soon

Let's Connect.

Open to research collaborations and AI/NLP projects. Feel free to reach out.