Deep Patel

AI/ML Engineer · LLM & RAG Specialist · GenAI Infrastructure

GitHub LinkedIn Resume

About Me

I’m an AI/ML Engineer with an M.S. in Computer Science from the University of Oklahoma, specializing in fine-tuning large language models (LLMs), Retrieval-Augmented Generation (RAG), and scalable ML systems. I build production-grade GenAI applications by combining LLM optimization (LoRA, QLoRA, 4-bit inference) with strong engineering foundations in PyTorch, Hugging Face, FastAPI, Docker, and AWS.

My work spans multi-document RAG assistants, multimodal reasoning systems, and high-performance backends for data-heavy applications—always with a focus on measurable impact: higher relevance, lower latency, and more reliable ML in production.

Experience

Machine Learning Engineer · Community Dreams Foundation

Remote, USA · Aug 2025 – Present
  • Managed the full ML lifecycle (requirements → modeling → evaluation), improving predictive performance by 22%.
  • Built Python + SQL pipelines for automated data ingestion and preprocessing, increasing dataset consistency.
  • Analyzed large community-impact datasets, extracting insights that improved model reliability by 30%.

Founding Machine Learning Engineer · TripRaft

Remote, USA · May 2025 – Aug 2025
  • Defined scalable backend architecture for itinerary planning, reducing API latency by 30% through query optimization and caching.
  • Shipped secure data schemas and REST APIs for collaboration, voting, and expense-sharing workflows.
  • Added ML personalization using vector search and ranking models, boosting trip discovery relevance by 18%.

AI/ML Engineer · Firenix Technologies

India · Mar 2021 – Jul 2023
  • Delivered PyTorch models for real-time classification, reducing prediction errors by 28% in production systems.
  • Accelerated inference by 40% using ONNX export and 8-bit optimization for edge deployments.
  • Developed end-to-end ML pipelines with preprocessing, inference services, and Dockerized FastAPI APIs.

ML Research Intern · IIIT Vadodara

India · Aug 2020 – Dec 2020
  • Achieved 82% accuracy on Alzheimer’s detection using a custom CNN trained on 492 PET scans.
  • Improved robustness using neuroimaging-specific data augmentations and cleaned preprocessing pipelines.

Education

M.S. in Computer Science

University of Oklahoma · Norman, OK · Aug 2023 – May 2025

B.Tech in Computer Science and Engineering

Indian Institute of Information Technology (IIIT) · Gujarat, India · Aug 2018 – May 2022

Featured Projects

NeuroDoc Project Screenshot
🧠 NeuroDoc: RAG Assistant

A multi-document RAG assistant using FAISS + BM25 hybrid retrieval, conversational memory, and citation-grounded responses with 18% improved precision.

RAGFAISSBM25FastAPI
View on GitHub
VisuaLens Project Screenshot
👀 VisuaLens: Multimodal LLM

A LLaVA-powered multimodal system for visual–text reasoning, dual-image comparison, and analysis with 4-bit quantized inference for efficient local deployment.

LLaVAVisual ReasoningQuantization
View on GitHub
Book Recommender Project Screenshot
📚 Semantic Book Recommender

Personalized book recommendations via semantic and emotional understanding using Sentence Transformers and an interactive Gradio interface.

Sentence TransformersGradio
View on GitHub
EcoAssist Project Screenshot
💬 EcoAssist: E-Commerce Chatbot

A domain-specific AI chatbot for e-commerce with fine-tuned GPT/LLaMA models, product-aware RAG, and real-time query resolution.

GPT/LLaMALoRAAWS Lambda
View on GitHub
EzyShop Project Screenshot
🛒 EzyShop: E-Commerce Platform

A full-stack e-commerce platform with secure auth, dynamic cart and order workflows, and optimized Python + SQL backend with 25% faster DB performance.

PythonSQLREST APIHTML/CSS
View on GitHub

Technical Skills

Programming & Data

PythonJavaSQLOOPData Structures & AlgorithmsRESTful APIsPandasNumPyPostgreSQLMySQLDynamoDB

ML & Deep Learning

Supervised & Unsupervised LearningRegressionClassification (Random Forest, XGBoost)Clustering (K-Means)Feature EngineeringPyTorchTensorFlowOpenCV

NLP & LLMs

Hugging Face TransformersRAG PipelinesLoRA/QLoRALangChainLangGraphVector DBs (FAISS, Pinecone)GPTLLaMAMistral

MLOps, Cloud & DevOps

AWS (S3, Lambda, EC2)DockerKubernetesFastAPICI/CDMLflowWeights & BiasesGitLinux CLI

Certifications

Generative AI with Large Language Models – DeepLearning.AI
Natural Language Processing with Transformers – Hugging Face
Building Transformer-Based NLP Applications – NVIDIA Deep Learning Institute
Fundamentals of Deep Learning – NVIDIA Deep Learning Institute

Let's Connect

Or, email me directly at pateldeep1842@gmail.com

Message Sent!