jayveersinh-raj Goto Github PK

followers: 16.0 following: 15.0 repos: 92.0 gists: 0.0

Name: Jayveersinh Raj

Type: User

Company: Innopolis

Bio: - I am a Machine learning / Deep learning practitioner. - I love Pokemons, AI, Cosmology (especially Quantum Mechanics), and Comedy. Get in touch if you too 😉

Location: India

Blog: github.com/Jayveersinh-Raj

Hi there

✨ I am Jayveersinh Raj, but you can call me Raj ✨

🔭 I’m a very curious and passionate ML/DL Engineer.
🌱 I’m mostly an NLP practitioner with experience in language models.
👯 I’m very passionate and up-to-date with the large language models field.
🤔 I’m also researching and working on Quantum Machine learning.
💬 Ask me anything about ML/DL, I am a very enthusiastic communicator
📫 How to reach me: 👆 Click on the icons below on which you want to reach out

⚡ Fun fact: I love pokemons, and I play Pokemon Go sometimes. Below is my pokemon team 👇, please contact me if you too love Pokemons

👆 My portfolio website:

⚙ My technological stack of experience

📖 Education

Bachelors 4th year, Majors: Data Science & AI

🏛️ University

Innopolis University, Tatarstan, Russia

💻 Some of the big projects undertaken and experience

Machine learning and AI

IOT-based anomaly detection
Console-based AI bot
Backend developer for IT department of Innopolis University
Cognizant virtual internship in AI (March 2023)
Timing attack based on Evolutionary algorithms (Proof of concept)
Spice Company scheduling optimization
Tennis tournament betting company forecasting for betting company in data mining
Ratings forecasting for airbnb and similar companies.

Natural Language

Diaster detection with NLP on news and tweets
Creating a natural language preprocessing pipeline for the autoML library Tpot that uses genetic programming.
Cross lingual zero-shot transfer (toxicity → done working on other derivatives) being trained on singular language data like English, can infer on any arbitrary language.
Gujarati Grammarly (Grammarly for low-resource language)
Vector space aligner for Sanskrit for there are nonexisting vector space aligner for this low-resource ancient language.
Using langchain for agents, prompt engineering, using vector databases to eliminate retraining of LLMs
Working knowledge of various alternatives of fine tuning LLMs.
Code generation/analysis research using LLMs with Advanced Engineering School (AES)

Computer Vision

Bounding box object detection using detectron2, and yolov8 while annotating from roboflow
Segmentation using detectron2, and yolov8 while annotating from roboflow
Digits recognition from triple-mnist custom datatset

Jayveersinh Raj's Projects

advancedquantum

My notes from Advanced Quantum Theory (taught by Tobias Osborne). This course covers the mathematical formalism and physics of many-particle quantum mechanics (in the framework of manifolds and configuration space), scattering theory, as well as introductory relativistic quantum mechanics.

authentification

baseline_gujarati_spell_checker

Baseline trial and conclusion of the gujarati spell checker based on Norvig's solution

baselinemodel_nb_sklearn_nlp

This is a Naive bayes classifier as a baseline for diaster analysis

bloom-lora

Due to restriction of LLaMA, we try to reimplement BLOOM-LoRA (much less restricted BLOOM license here https://huggingface.co/spaces/bigscience/license) using Alpaca-LoRA and Alpaca_data_cleaned.json

boston_housing_neural_regression

classification_decision_boundary_function

code_generation_gpt2

Fine tuning a gpt2 model for code generation/completion. This is the work under Advanced Engineering School (AES) Innopolis University. Data and model is not open source, the prototyping is for learning purposes.

cognizant-virtual-internship

This repository contains the work in the AI engineer Cognizant virtual training and internship program from forage

comparch

console_bots_cpp

control-theory-slides-spring-2022

cross-lingual-zero-shot-transfer

A cross lingual toxicity detection model that works for over 100 languages. Powered the mighty XLM-R model, the model performance is state of the art.

cross_lingual_abuse_detection

In this work we applied multilingual zero-shot transfer concept for the task of toxic comments detection. This concept allows a model trained only on a single-language dataset to work in arbitrary language, even low-resource.

cross_lingual_llm_for_everything

Multilingual spellings correction and Question Answering Large Language Models

cs224u

Code for Stanford CS224u

cvessentials

Tutorial Series (60 hour course): Essentials of computer vision

data_cleaning_infosys_dataset

databases_course_2022

detectron_and_yolov8

This is the assignment 1 of object detection and comparison between faster RCNN detectron2, and Yolov8. It is provided under the course of Advanced machine learning at Innopolis University. Roboflow was used for annotations.

differntialeq_analyticalmethods_code

dijktra_algorithm_c-

distributed_systems_programming

This is the collection of labs on the Distributed Systems and networks programming course offered at Innopolis University. I will keep it updated as I move forward.