Available for Innovation

Chetan Valluru

Data Scientist • LLM Engineer • ML Specialist

Building production LLM agents, RAG systems, and on-device ML solutions. Specialized in generative AI, NLP, and MLOps — reducing costs 60%, accelerating insights 80%, achieving 99.9% accuracy. Expert in PyTorch, Transformers, fine-tuning (LoRA), quantization, and privacy-preserving deployment.

500+
Connections

Chetan Valluru

Focused on building production-grade LLM agents, scalable RAG systems, and advanced generative AI frameworks.

View Profile
Python PyTorch
View Repositories

About Me & Education

lightbulb My Mission

I build production AI systems that solve real problems. From architecting NL-to-SQL platforms with LLM-powered NLP to deploying privacy-preserving on-device ML, I focus on shipping reliable, optimized solutions. My work has reduced data tickets 30%, accelerated insights 80%, and enabled self-serve analytics for cross-functional teams. I specialize in LLMs, RAG, fine-tuning, quantization, and MLOps — always balancing performance, cost, and security.

school Education

Rutgers University — M.S. Data Science

GPA: 3.65 Sep 2023 – May 2025

Focus: LLMs, Generative AI, MLOps, System Design

VNR VJIET — B.Tech Electronics & Communication

GPA: 3.3 Aug 2019 – May 2023

Specialization: AI & Machine Learning

Distinctions

article

IEEE Research Publication

Multi-Classification of Respiratory Diseases using Deep Learning

2023 ICSCSS • DOI: 10.1109/ICSCSS57650.2023.10169597

emoji_events

HackRU Fall 2024 — 1st Place

NJ Transit Track Winner

ML delay prediction (82% accuracy) + RAG chatbot serving 15K+ queries.

military_tech

Global Youth Scientist

GOLD Award

Recognized for applied AI research and community impact in international STEM competition.

Experience

analytics
August 2025 – Present

Data Scientist

JerseySTEM, NJ

  • Architected production NL-to-SQL platform using LLM-powered NLP, LangChain, and schema-aware RAG; reduced data tickets 30%, accelerated insights 80%, enabled self-serve analytics for 8+ teams.
  • Engineered MLOps pipeline for LLM deployment with NLP evaluation harness (faithfulness, semantic similarity, precision/recall) and automated testing; achieved 95% grounded responses via optimized LLM inference.
  • Implemented LLM security: NLP-based PII detection, RBAC, audit logging, prompt injection mitigation; reduced LLM hallucinations 40%, maintained 99.8% uptime Q4 2025.
flight_takeoff
Oct 2024

Data Science Virtual Intern

British Airways via Forage

Conducted NLP sentiment analysis on 3,949+ reviews; built predictive models to map buying patterns and proposed experiments for ~20% campaign ROI lift.

school
Sep 2023 – May 2024

Machine Learning & Statistics Grader

Rutgers University

Reviewed 60+ weekly assignments on statistical inference and ML. Focused feedback on reproducibility, metrics validation, and model-data fit.

science
Aug 2021 – May 2023

AI/ML Research Assistant

VNR VJIET, India

  • Fine-tuned multilingual NLP models (mBERT, IndicBERT) using LoRA and 8-bit quantization on PyTorch; improved access for 50,000+ farmers, reduced language barriers 40%, published reproducible benchmarks supporting government agricultural programs.
  • Built ETL data pipelines (Python, Scrapy, SQL) with validation, drift detection, lineage tracking; generated 2M+ documents accelerating regional AI research 60%.
  • Designed eval framework (BLEU, ROUGE, F1) with chaos tests and fallbacks; maintained 99.5% uptime across 6-month production pilot.

Featured Projects

Generative AI NLP

Taylor: AI Career Stylist

RAG-powered LLM app automating resume/cover letter creation for 250+ users. 50% faster application processing, <3s latency, serverless architecture with personalized outputs.

👁️
RAG / ML

GemmA.I: Vision Assistant

iOS accessibility app for blind/low-vision users with real-time scene understanding via MediaPipe Gemini. 100% on-device privacy, <500ms inference (INT8).

Fine-Tuning

BetterLlama

Fine-tuned LLaMA 3.2 with LoRA/8-bit quantization. 1.8× faster inference, 99.9% MMLU accuracy, 4× compression, 60% cost reduction. Integrated chain-of-thought and LLM Guardrails.

Transformers

ChessGPT

Custom tokenizer for move generation (~30 legal moves/prompt) with compact model footprint. GPT-based architecture for chess move prediction.

XGBoost Agents

NJ Transit Assistant

1st Place HackRU. XGBoost delay prediction (500K+ records, 82% accuracy) + schema-aware RAG chatbot. 70% faster FAQ resolution, 15K+ queries processed.

Technical Stack

Core competencies across the AI development lifecycle

LLMs
Generative AI
NLP
Transformers
RAG
Fine-Tuning
LoRA
Quantization
Prompt Engineering
PyTorch
TensorFlow
HuggingFace
LangChain
Scikit-learn
XGBoost
MLOps
Docker
CI/CD
GCP
Vector DBs
PostgreSQL
MongoDB
Python
Swift
SQL
JavaScript

Let's build the future of AI together.

Currently open to research collaborations and engineering opportunities in generative AI and large-scale data systems. Based in United States.

LinkedIn GitHub