Hi, I'm Aryan Patodiya

AIML Developer

Building impactful, scalable AI solutions with expertise in machine learning, deep learning, and cloud infrastructure.

View Projects Contact Me
Aryan Patodiya

About Me

Aryan Patodiya
1.5+ Years
Experience

Graduate Computer Science student with ~1.5 years of hands-on experience through research, internships, and startup work in machine learning, backend development, and cloud infrastructure.

Built end-to-end ML pipelines at ISRO, deployed real-time systems using AWS and Azure, and developed LLM-based applications for document search and forecasting. Skilled in Python, C++, deep learning frameworks, and MLOps tools.

Eager to contribute to impactful, scalable AI solutions that solve real-world problems and drive innovation.

Education

MS in Computer Science

Experience

ISRO, Raven Technolabs, NanoTech Technologies

Projects

ML/AI Applications

Download Resume

Skills & Expertise

Programming

Python

Python

Primary language for ML/AI development

Java

Java

Backend development & enterprise applications

C++

C++

Performance-critical applications & algorithms

SQL

SQL

Database querying & management

JavaScript

JavaScript

Web development & interactive applications

TypeScript

TypeScript

Type-safe JavaScript development

Machine Learning & Deep Learning

PyTorch

PyTorch

Deep learning research & development

TensorFlow

TensorFlow

Production ML systems & deployment

OpenCV

OpenCV

Computer vision applications

Transformers

Transformers

NLP models & applications

GPT

GPT

Large language model applications

CNNs

CNNs

Computer vision & image processing

RNNs

RNNs

Sequence modeling & time series

LLMs

LLMs

Large language model development

RL

Reinforcement Learning

Agent-based learning & optimization

CUDA

CUDA

GPU acceleration for deep learning

Model Serving & Optimization

Triton

Triton Inference Server

High-performance model serving

vLLM

vLLM

High-throughput LLM inference

ONNX

ONNX

Model interoperability & exchange

TensorRT

TensorRT

High-performance deep learning inference

TorchScript

TorchScript

PyTorch model optimization

Cloud & MLOps

AWS

AWS

Cloud infrastructure & services

Azure

Azure

Microsoft cloud platform

Docker

Docker

Containerization & deployment

Kubernetes

Kubernetes

Container orchestration

Terraform

Terraform

Infrastructure as code

CI/CD

CI/CD

Continuous integration & deployment

MLflow

MLflow

ML lifecycle management

Data Engineering & Big Data

Kafka

Apache Kafka

Distributed event streaming

Hadoop

Hadoop

Distributed data processing

MongoDB

MongoDB

NoSQL database

Redis

Redis

In-memory data store

MySQL

MySQL

Relational database

PostgreSQL

PostgreSQL

Advanced relational database

Software Development

OOP

Object-Oriented Programming

Software design & architecture

Design Patterns

Design Patterns

Reusable software solutions

REST & GraphQL

REST & GraphQL APIs

API design & implementation

Multi-threading

Multi-threading

Concurrent programming

Agile

Agile Methodologies

Scrum, Kanban, & project management

Work Experience

SAC-Indian Space Research Organization

ML Intern – Hydrological Modeling & Forecasting

Dec 2022 - Apr 2023

  • Built a forecasting pipeline using HMMs, Markov Chains, and LDA to model rainfall patterns; improved prediction accuracy by ~20%.
  • Processed and cleaned hundreds of GBs of satellite data using Pandas and AWS S3, reducing prep time by ~30%.
  • Integrated results into flood risk tools, helping improve early warning reliability across pilot regions.

Raven Technolabs

ML Engineering Intern – Cloud Infrastructure & Model Deployment

May 2022 - July 2022

  • Developed backend microservices in Spring Boot and Node.js for real-time ML model deployment.
  • Created REST/GraphQL APIs to support model inference; improved latency by ~25%.
  • Automated CI/CD pipelines using GitHub Actions and AWS CodeDeploy, reducing manual deployments by 50%.

Nanotech Technologies

Cofounder & Lead Software Engineer – Scalable Systems & Data Engineering

Mar 2019 - Nov 2021

  • Led a 14-member team to build scalable edge-to-cloud data pipelines for industrial automation.
  • Optimized backend systems via low-level code refactoring and DB tuning, improving performance by ~20%.
  • Built infrastructure for real-time monitoring and enabled future ML use cases like predictive maintenance.

California State University

Brain-Computer Interface Research Assistant

May 2025 - August 2025

  • Designed a scalable EEG preprocessing pipeline (MATLAB, Python) processing 64k+ multi-channel trials, cutting preprocessing time by 40%.
  • Built optimized CNN architectures with temporal–spatial convolutions, SE blocks, and GELU activations, achieving 98.7% balanced accuracy and AUC 0.9997.
  • Implemented a robust evaluation framework (test-time voting, stratified CV, augmentation), improving model robustness by 25% and leading to international conference acceptance.

California State University

SI Leader - Instructional Student Assistant

August 2025 - Present

  • Facilitated peer-led learning sessions for 20+ undergraduate students, reinforcing lecture topics through interactive activities such as concept mapping, group problem-solving, and coding challenges.
  • Designed and delivered structured lesson plans (icebreakers, warm-ups, main activities, and closing) that enhanced comprehension of core computer science concepts including control structures, processes, and threads.
  • Cultivated a collaborative learning environment by guiding discussions with probing questions, encouraging peer-to-peer explanations, and supporting diverse student learning styles with adaptive strategies.

Projects

Explore my latest work in AI/ML, Reinforcement Learning, and Fullstack Development

MarketPulse: Real-Time Stock Trend Predictor
GitHub

MarketPulse: Real-Time Stock Trend Predictor

LSTM GRU Streamlit AWS

Designed LSTM/GRU models achieving ~80% directional accuracy on OHLCV data. Deployed using TensorFlow Serving on AWS with vector embeddings, reducing query latency by 20%.

DocuQuery: Fullstack Semantic Search Engine
GitHub

DocuQuery: Fullstack Semantic Search Engine

LLaMA-2 LangChain FAISS Streamlit

Built a context-aware search system using LLaMA-2, LangChain, and FAISS. Integrated OpenAI embeddings to boost relevance by ~30% and deployed as a Dockerized Streamlit microservice.

CollabBoard

CollabBoard – Real-Time Team Collaboration Platform

React.js Node.js WebSockets MongoDB AWS

Developed a production-ready collaboration app supporting 1,000+ concurrent users with real-time WebSocket updates, AWS CI/CD deployment, and authentication-based role control.

Personal Finance Management Platform

Personal Finance Management Platform

React.js Node.js PostgreSQL D3.js Chart.js

Built a full-stack finance tracking system using React.js, Node.js, and PostgreSQL. Integrated Plaid API for secure bank transactions and real-time analytics visualizations.

Real-Time Object Detection using YOLOv5

Real-Time Object Detection using YOLOv5

YOLOv5 TensorRT OpenCV COCO

Implemented YOLOv5 on COCO dataset achieving 85% mAP. Optimized inference by 30% using TensorRT acceleration and integrated OpenCV for 30 FPS real-time tracking.

CDPC Management

CDPC Management System

React.js AWS Lambda API Gateway

Developed a React-based career management system that reduced manual workload by 50% and improved efficiency by 30%. Integrated AWS Lambda & API Gateway for 99.9% uptime.

Web 3.0 App

Web 3.0 App – Decentralized NFT Marketplace

React.js Solidity AWS

Engineered a decentralized NFT marketplace using Solidity and AWS, handling $10,000+ in digital assets. Designed a gas-free smart contract system to eliminate fees entirely.

Deep Learning for Image Classification

Deep Learning for Image Classification

CNN ResNet TensorFlow

Developed CNN and ResNet architectures for MNIST and CIFAR-10 datasets, increasing accuracy from 85% to 98%. Enhanced performance with data augmentation and hyperparameter tuning.

NdLinear Benchmarking for Reinforcement Learning

NdLinear Benchmarking for Reinforcement Learning

NdLinear REINFORCE CartPole LunarLander

Benchmarked NdLinear layers against nn.Linear in RL tasks. Achieved faster convergence, higher stability, and smoother reward curves across CartPole-v1 and LunarLander-v2 environments.

TimeNet

TimeNet: Sequence Forecasting with RNNs

RNN TensorFlow Forecasting

Developed RNN models for rainfall and energy forecasting, outperforming ARIMA by 15–20%. Optimized data pipeline and reduced training time by 20%.

Article Summarizer

Article Summarizer

NLP Summarization Sentiment Analysis

Developed an NLP-based summarizer reducing text length by 80% while maintaining 95% accuracy in sentiment analysis. Used advanced algorithms for text compression and semantic scoring.

SkyDefender 3D Game

SkyDefender: 3D Shooter & Duck Hunt Game

C++ OpenGL GLUT SOIL irrKlang

Developed a 3D OpenGL-based duck shooting game with parallax environments, pathfinding drones, and collision detection. Achieved ~60 FPS through optimized audio-visual subsystems.

Optimizing RAG with Reinforcement Learning

Optimizing Retrieval-Augmented Generation (RAG) with Reinforcement Learning

PPO RAG SQuAD/NQ Evaluation

Designed a custom PPO-based RL framework improving RAG retrieval relevance by 20–25%. Integrated BLEU and Exact Match metrics for robust evaluation and generalization.

Education

California State University, Fresno

Master of Science in Computer Science

Jan 2024 - Present | GPA: 3.9/4.0

Relevant Coursework: Advanced Machine Learning, Deep Learning, Natural Language Processing, Computer Vision, Reinforcement Learning, Distributed Systems

Charusat University

Bachelor of Technology in Computer Science and Engineering

July 2019 - Apr 2023 | GPA: 3.82/4.0

Relevant Coursework: Data Structures and Algorithms, Operating Systems, Database Management Systems, Computer Networks, Artificial Intelligence, Machine Learning

Contact Me

Let's discuss how we can work together

Email

aryanpatodiya018@gmail.com

Phone

+1 (925)-918-1058

Location

Fresno, CA