HOME ABOUT PROJECTS WORK CONTACT

Welcome to my Domain!

I'm
Jay Soni

Data Scientist & Analyst.

About Me


I am a Data Scientist with 2+ years of experience and proven knack to improve business decisions by transforming raw, unstructured data into actionable insights that can optimize business processes and drive peak performance for the enterprise.

I am here to assist you to:
👉Ask questions to make data driven decisions.
👉To prepare the data for exploration.
👉To help process your dirty data to clean.
👉To answer all your queries and questions by analyzing the data.
👉By Sharing the insights through the visualization and reports.
👉And, Finally, helping to deeply understand what the insights are conveying & suggesting you how to ACT for it.

My Skills

Python

Machine Learning

Deep Learning

Tableau


OTHER SKILLS
C++, SQL, Statistical testing, Financial Modeling, Time series analysis, Probability, Risk modeling, MS Excel

Projects



SuperStore Sales Analysis

Norway

Conducted comprehensive sales data analysis, leveraging techniques such as data visualization, statistical modeling, and data mining and designed an interactive sales dashboard using Tableau to facilitate data-driven decision-making.

Access Here

Financial Market Sentiment Analysis Using Twitter Data

Norway

By leveraging the power of machine learning and natural language processing, I've developed a model that can help you see through the fog of Twitter noise. Using a dataset of tweets related to financial markets, I've trained a Support Vector Machine (SVM) model to analyze sentiment of stock market.

Access Here

Bitcoin Beta Forecast Model

Norway

Developed a machine learning model to predict 1 week ROI of Bitcoin prices by calculating least MAE from variety of algorithms like SVR, random forest, LSTM, stochastic gradient regressor and rolling forecast. Achieved 2.25% MAE by testing 5 different techniques and 64.3% accuracy in ROI Prediction.

Access Here

Retail Store Propensity Model

Norway

Classification models including logistic regression and random forest are introduced along with their diagnostic methods and tuning processes. Variable importance analysis is highlighted following with comparisons of two models.

Access Here

WORK EXPERIENCE


Freelance Data Scientist & Analyst

Aug 2023 - Present

- Built a Customer Behavioral Progression propensity model for a retail store using XGBOOST for multi-class classification using insights from the last decision tree in the XGBOOST ensemble.
- Contributed towards machine learning, customer segmentation analytics, statistical modeling to drive marketing strategies by creating interactive dashboards and reports using data.Table
- Automated Excel reports using R Programming and k-means algorithm in R for optimal customer segmentation.
- Trained a state-of-the-art Large Language Model (LLM) as a Machine Learning Engineer in facilitating Reinforcement Learning via Human Feedback (RLHF) to fine-tune Al models with an accuracy of 88.5%.
- Engineered predictive to analyze the air quality in Nairobi and improved forecast accuracy by 21%.


Technology Journalist

Electronics For You Group

July 2022 - April 2023

- Conducted in-depth market research on India's technology sector, integrating numerical and visual metrics using MS Excel and Power BI; delivered a comprehensive report that projected over 20% growth potential in emerging industries by 2030.
- Successfully executed qualitative research and descriptive analysis to revamp an ‘online B2B Electronics directory’ with over 10,000 companies listed across 15 categories .
- Published news articles and reports based on interviews, innovative researches and tech startups in India.


Embedded Systems Intern

Myth Interactives

March 2022 - June 2022

Revamped a faulty system integrating multiple sensors and protocols with Raspberry Pi (Linux OS).


Student Researcher

V.V.P Engineering College

September 2021 - February 2022

- Designed a hardware system to acquire Brain signal using TMS320VC5509A digital processor and implemented a Low Pass Filter to improve signal strength for better signal recording.
- Constructed a data model on MATLAB to detect and filter spike(a signal feature), caused due to eye movements, from brain signals in order to identify abnormalities in brain function, achieved 89% accuracy.

EDUCATION


MBA - Data Science & Analytics

Manipal University Jaipur

Aug 2023 - Aug 2025 (Expected)

RELAVANT COURSEWORK:
Management Information Systems, Project Management, Advanced Machine Learning For Finance, Insurance & Risk Management, Business Intellignece & Visualization, Business Analytics, Statistics, Security Analysis & Portfolio Management

Bachelor of Engineering -
Electronics & Communication

V.V.P Engineering College - Gujarat Technological University (GTU)

Aug 2019 - May 2023

Contact Me


Bhuj, Gujarat, India

Email: jayamitsoni10126@gmail.com