I'm Gitesh
AI/ML Engineer
& Researcher
Building intelligent systems that understand, learn, and adapt
Specialized in deep learning, computer vision, and natural language processing. I design and implement cutting-edge AI solutions that push the boundaries of what's possible with machine intelligence.
|Featured Projects
Keras Vision Models
Keras Vision Models Library (KVMM), a comprehensive library of vision models with pre-trained weights entirely in Keras 3. KVMM supports a wide range of computer vision tasks including segmentation, classification, and vision-language modeling.
Chat with PDFs using LLMs
A AI agent built using LlamaIndex that leverages RAG (Retrieval-Augmented Generation) architecture along with Llama-2 and sentence transformers to create an efficient search and summarization tool for PDF documents. This tool allows users to query information from PDF files using natural language and obtain relevant answers or summaries.
Image Super Resolution Using EDSR and SRGAN
Implementation of EDSR and SRGAN reseach papers for image super-resolution tasks. This project demonstrates how to enhance the resolution of images using deep learning techniques, providing a practical application of convolutional neural networks in computer vision.
YOLOv5 & YOLOv8 Implementation in Keras 3
A complete from-scratch implementation of YOLOv5 and YOLOv8 object detection models in Keras 3, including utilities for converting pretrained PyTorch weights for seamless usage in Keras.
Core Expertise
Comprehensive AI/ML expertise spanning from traditional machine learning to cutting-edge deep learning architectures and scalable data science solutions.
- Machine Learning & Deep Learning
Advanced expertise in ML algorithms and neural network architectures. Specialized in computer vision, NLP, LLMs, attention mechanisms, and transformer models.
Experience: 3+ years
KVMM: Keras Vision Models library
Proficiency88%TensorFlowPyTorchKerasScikit-learnTransformersObject DetectionImage Segmentation- Agentic AI
Building intelligent AI agents and conversational systems. Expert in RAG architectures, LLM integration, and document processing with advanced NLP capabilities.
Experience: 2+ years
Chat with PDFs using LLMs
Proficiency75%LangChainHuggingFaceGroqLlamaIndexRAG- Full-Stack Development
End-to-end web application development with modern frameworks. Expert in building scalable applications with cloud infrastructure and security best practices.
Experience: 1 year
Production deployments at Fiserv
Proficiency80%React.jsJavaScriptTypescriptAWSGraphQL
Professional Experience
A journey through cutting-edge AI/ML roles, from research breakthroughs to production systems at scale
Technology Analyst - SDE
Deep Learning Intern
Open Source
Actively contributing to the open-source community with a focus on AI/ML tools and libraries.
Keras
Core contributor to Keras core library, developing image processing layers and mathematical operations.
Key Contributions:
- •Implemented RandomGrayscale, RandomHue, and Equalization layers
- •Added mathematical operations (unravel_index, diagflat, inner)
- •Developed PSNR API for image quality assessment
Keras.io Documentation
Contributed comprehensive documentation and tutorials to the official Keras documentation site.
Key Contributions:
- •Created TensorFlow Decision Forest integration docs
- •Developed YOLOv8 training implementation tutorial
KerasCV
Contributed computer vision models and loss functions to the Keras Computer Vision library.
Key Contributions:
- •Implemented CIoU loss function for YOLOv8
- •Implemented ConvMixer architecture
- •Contributed to the migration process to Keras 3.
- •Assisted in migrating models to the functional-subclassing API.
HuggingFace Transformers
Contributed optimizations for vision transformer implementations, documentation improvements, and enhanced testing coverage.
Key Contributions:
- •Implemented comprehensive tokenization tests for LED model architecture
- •Enhanced type safety by adding PyTorch-specific type hints
- •Added missing tokenization test coverage for ELECTRA model