Software Development Engineer - NLP, LLMs, RAG Systems

PRAVEEN KUMAR PALABOYINA

Machine Learning Engineer NLP LLMs RAG Systems

**********************@*****.*** +91-911******* linkedin.com/in/praveenkumarpalaboyina

github.com/praveenkumar911

Designed and implemented end-to-end OCR pipeline using PyTesseract with custom preprocessing, achieving 95% text extraction accuracy and reducing manual data entry workload by 70% Built full-stack Shipping Assistant application (React + Flask) with intelligent document processing, improving operational workflow efficiency by 40%

Transformers, Large Language Models (LLMs), RAG Systems, PyTorch, TensorFlow, HuggingFace, NLP, OCR, Deep Learning

FAISS, MB25, Vector Databases, Model Pre-Training, FineTuning MLOPs. Python, JavaScript

Node.js, Express.js, Flask, FastAPI, RESTful APIs

AWS, Docker, CI/CD Pipelines, NGINX

LangChain, PyTesseract, Grafana, Kibana, ELK Stack, Scikit-learn, Pandas, NumPy Architected and deployed production-grade RAG pipelines using FAISS vector databases and Transformer models, achieving 35% improvement in retrieval relevance and answer quality Engineered AI-powered educational platform serving 5,000+ rural learners with personalized learning experiences and real-time content recommendations

Developed automated analytics system for classroom engagement tracking, providing actionable insights for educators Managed distributed ML infrastructure leveraging Docker containerization, NGINX load balancing, and ELK Stack monitoring, maintaining 99.9% uptime across production environments Led CI/CD pipeline implementation for automated model deployment, reducing deployment time by 60% and ensuring seamless production updates

Optimized inference pipelines through model quantization and batching strategies, reducing latency by 45% while maintaining model performance

Mentored team of 4 junior engineers and interns on ML best practices, code reviews, and production deployment workflows TECHNICAL SKILLS

PROFESSIONAL SUMMARY

PROFESSIONAL EXPERIENCE

Results-driven Machine Learning Engineer with 2.5+ years of experience specializing in Large Language Models, Natural Language Processing, and Retrieval-Augmented Generation systems. Proven expertise in designing and deploying production-grade AI solutions using Transformers, PyTorch, and scalable ML infrastructure. Successfully built AI-powered education platforms serving thousands of users and developed efficient OCR pipelines. Published researcher (IEEE TALE) with strong foundation in end-to-end ML product development, optimization, and deployment at scale. Programming:

Backend & APIs:

Cloud & DevOps:

ML Tools & Libs:

ML Infrastructure:

Machine Learning:

Software Engineer (FULL STACK MERN)

RCTS-IIITH

Software Development Engineer (AI/ML)

RCTS-IIITH

May 2023 – November 2024

November 2024 – Present

EDUCATION

PUBLICATIONS

KEY PROJECTS

ACHIEVEMENTS & LEADERSHIP

Multilingual IVR System

Automated Classroom Analytics Suite

RAG-Based Environmental Learning Agent

Multilingual Language Model (English–Telugu–Konkani) Master's by Research in Computer Science (Part-Time) Bachelor of Technology in Electronics & Communication Engineering

— IIIT Hyderabad

— KIET 2019 – 2023

2025 – Present

— Toastmasters International (2021)

— Global Coding Club KIET (2020–2023)

"AI Framework for Scalable Automated Continuous Formative Assessment" — IEEE TALE Conference Built Retrieval-Augmented Generation system using FAISS vector store and Transformer models for K-12 education content Improved factual accuracy by 40% through domain-specific retrieval and answer grounding in curated academic content Implemented hybrid search combining dense embeddings and keyword matching for enhanced retrieval precision Developed NLP-based sentiment analysis and engagement tracking tools using BERT models to analyze student feedback and learning patterns

Implemented automated data validation and quality checks, reducing data errors by 55% across production workflows

— KIET Toastmasters Club

— IEEE TALE Conference

Built real-time computer vision pipeline for facial recognition-based attendance and attention monitoring using OpenCV and deep learning models

Designed distributed microservices architecture to handle parallel processing of multiple video streams with low latency Achieved 94% accuracy in attention detection and 98% accuracy in facial recognition across diverse lighting conditions Created speech-driven Interactive Voice Response system answering Physics and Biology queries in Telugu using speech recognition and TTS

Reduced model inference time from 40s to under 15s through intelligent caching, model optimization, and response pre-computation Integrated with telephony systems to provide accessible educational content to rural students without internet access Developed compact multilingual transformer-based LLM capable of sentence completion, translation, and sarcasm detection across three languages

Optimized model inference using ONNX runtime and INT8 quantization, reducing model size by 75% and achieving real-time response latency under 200ms

Implemented custom tokenization strategy to handle code-mixed text and improve cross-lingual understanding

• Co-Author:

Area Champion

Student Leader

Vice-President Membership

Published Researcher

Contact this candidate