Experience

Projects

Stock Price prediction

• In this project, I developed a predictive model to analyze and forecast stock prices, helping inform investment decisions and assess stock profitability across various time frames. To identify the most suitable algorithm, I initially tested Multivariate Linear Regresion and uto Regreivev Integrated oving Average(ARIMA) models. However, these approaches lacked accuracy due to the non-linear nature of stock data and their limitations in predicting turn points.
• After researching non-linear models, I implemented a Long Short-Term Memory (LSTM) neural network, which is particularly effective for time-series problems due to its ability to learn long-predictions, I cross-verified the results by designing and testing multiple models that analyzed data on monthly, quarterly, and yearly intervals. This multi-scale approach enabled more robust prdictions by capturing patterns across different time frames.
• Key components of this project included data preprocessing, feature engineering, hyperparameter tuning, and performance evaluation. The LSTM model ultimately outperformed traditional models, yielding more accurate and actionable insights. This project reflects my expertise in time-series analysis, machine learning, and deep learning, as well as my commitment to rigorous model Validation.

Family Asset Management

• In this project, I and my team of 4 people developed a Family Asset Management System in ReactJS, to simplify and securely organize various family assets. The system enables users to track, categorize, and analyze assets across multiple categories, including real estate, investments, personal valuables, and savings. My goals providing valuable insights into assets distribution and growth over time.
• To ensure data security, I encorporated robust safeguards to protect sensitive Information. Key features include automated data aggregation, financial performance tracking, and customizable reporting. By centralizing asset information, this tool aids in financial planning. This project demonstrates my abilities in data organization, financial analysis, and secure data management, ultimately contributing to a more efficient and informed approach to family asset management.

Object Classification

• In this project, I developed a machine learning model to perform object classification, leveraging deep learning techniques to enhance the accuracy and efficiency of identifying and categorizing objects within images. My obective was to create a model that could effectively generalize across a diverse dataset and deliver reliable predictions in various environments.
• To achieve this, I used Transfer Learning to speed up model training and improve performance by utilizing pre-trained netorks. I explored and fine-tuned multiple architectures, carefully selecting the best-performing model optimization to handle challenges such as variations in lighting, orientation, and background. For improving accuracy of the model we trained the pre-trained model on the augmented data
• This project honed my skills in image processing, feature engineering, and machine learning, showcasing my ability to create robust solutions for complex classification tasks and solve real-world challenges in object detection and recognition.

Speech Sentiment Analysis

• In this project, I in a group of 3 people, we built a Speech Sentiment Analysis model designed to evaluate and classify emotions within spoken audio data. Our goal was to create a system capable of accurately identifying sentiment—such as positive, negative, or neutral—from speech, with applications in customer service, feedback analysis, and emotion moitoring.
• To develop the model, We worked with the both audio preprocessing and feature extraction, focusing on capturing key vocal attributes like pitch, tone, an rhythm. I experimented with various machine learning algorithms and deep learning architectures, ultimately selecting a model that could best capture emotional cues from vocal intonations. We also performed rigorous testing across different languages and accents to ensure that the model generalized well across diverse speech patterns.
• In this project I got to showcase my skills in natural language processing(NLP), audio processing, and machine learning, highlighting my ability to create data-driven solutions for nuanced, real-world challenges in emotion recognition and sentiment analysis.

Generation and Optimization of time-table using Genetic Algorithm

• In this project, I in a team of 4 developed a timetable generation system that uses a genetic algorithm to create optimized schedules. The goal was to automate the scheduling process, reducing conflicts and ensuring efficient resource alloction for classes, instructors, and rooms. By simulating natural selection and evolving potential schedules, the genetic algorithm iteratively improved timetable quality, handling constraints such as overlapping classes, instructor availability, and room capacity.
• I designed the system to balance efficiency and flexibility, enabling it to adapt to a range of scheduling scenrios. This involved implementing key genetic algorithm operations, including selection, crossover, and mutation, to find optimal solutions. The project highlighted my skills in algorithm design, optimization, and problem-solving, demonstrating the effectiveness of evolutionary algorithms in handling complex scheduling tasks.
• We also published a paper on this project, showcasing our research and presentation skills.

Tweepy

• For this project, I used Tweepy, a Python library for accessing the Twitter API, to collect process, and analyze Twitter data. Our goal was to gather insights on trending topics, public sentiment, user engagement, leveraging Twitters real-time data to identify patterns and trends in social media behavior. We designed the system to fetch and filter tweets based on keywords, hashtags, an use mentions, ensuring targeted data collection.
• After gathering the data, We applied natural language processing techniques to clean and analyze the text, enabling sentiment analysis and topic modeling. This project strengthened my skills in data extraction, sentiment analysis, and social media analytics, and showcased my ability to work with APIs to derive actionable insights from unstructured social data.

Property Management

• In this project, I designed a Property Management System in a team of 4, to simplify and enhance the management of real estate properties. Using Java and Java-servelets for backend development and MySQQL for database management, I created a robust system that allows property owners and managers to track lease information, tenant details, maintenance schedules, and financials. The fronten was developed with JavaScript, HTML, and CSS to deliver a responsive, user-friendly interface, while we used XAMPP server for legacy deployment.
• The system's feature include automated reminders for lease renewals, financial reporting, and maintenance alerts, significantly reducing administrative workload and enhancing operational efficiency. This project highlighted my skills in backend and frontend development, database management, and deployment, resulting in a streamlined, data-driven property management solution.

Maharashtra Tourism

• For this project, I with my team-mate analyzed tourism data in Maharashtra to uncover patterns and trends that could support tourism growth and improve visitor experiences. Leveraging Data Mining skills to extract tourism data from various travel blogs and websites, we conducted in-depth analysis on factors such as tourist volume, popular destinations, seasonal trends, and demographic insights. I used Tableau to create interactive dashboards, enabling stakeholders to gain clear, actionable insights.
• Through this project, We developed visual reports and dashboards that revealed key influences on tourist behavior, such as peak travel seasons and attraction popularity. This project demonstrated my strength in data analysis, visualization, and reporting, and underscored my ability to use data science tools to support strategic decision-making in tourism.

Techincal Skills

• Development Skills: Software Development, Front-end Development, Back-end Development, HTML(5), CSS(3), Bootstrap, JSP, Servlet, ReactJS, Java, JavaScript, Php
• Data Science Skills: Data Analytics, Data Visualization(PowerBI, Tableau), Data Mining, Text Mining, Numpy, Pandas, Tensorflow, Matplotlib, Torch, Scipy
• Data-Bases: MySQL, MongoDB, NoSQL
• Programming Languages: C, C++, Python, Java

Certificates:

• Introduction to JavaScript - Microsoft Technology Associate
• Introduction to Artificial Intelligence (AI) - IBM via CourseEra
• Data-driven Decision Making - PWC via CourseEra
• Business Analytics - Wharton University of Pennsylvania via CourseEra
• Customer Analytics - Wharton University of Pennsylvania via CourseEra
• Operations Analytics - Wharton University of Pennsylvania via CourseEra
• People Analytics - Wharton University of Pennsylvania via CourseEra
• Accounting Analytics - Wharton University of Pennsylvania via CourseEra
• Google Analytics(Beginners) - Google Analytics Academy
• Advanced Google Analytics - Google Analytics Academy

Conferences:

Attended PyCon India in both 2018 and 2019, where prestiogious companies presented their projects in Python and data science. Also, Received insights into the latest advancements in Machine Learning and Artificial Intelligence technologies prevalent during those years.

Hackathons:

• HackUTA 2022, participated and emerged victorious in the challenge by MLH, Best use of Velo by WIX. Leveraged the Velo by WIX platform to enhance the User Interface for displaying Statefarm's insurance policies. Additionally, we created an improved interface connecting agents and clients based on their locations, employing Twilio to ensure a secure and protected communication channel between the two. The project was developed using agile methodologies to maintain the quality of the development process.
• Axxess Hackathon 2023, secured the 3rd place in the overall hackathon standings. Our project aimed to enhance hygiene and incorporate technology in the medical industry. We devised an application enabling gesture-based control, ensuring no direct contact between the operating surgeon and the system. The implementation involved utilizing OpenCV, MediaPipe, ReactJs, NodeJs, and MySQL, contributing to our success in achieving this innovation.
• HackSMU V, engaged in a competition and secured the award for effectively utilizing MongoDB Atlas. Developed an application named Hive, with the aim of simplifying the process of donating and sharing unneeded belongings. Our platform facilitates seamless sharing, fostering a stronger sense of community collaboration.