• Spearheaded end-to-end ML project integration with MLflow and AWS GitHub Actions for CI/CD.
• Tailored user interactive app deployment on AWS ECR, employing Docker for seamless integration.
• Amplified project management efficiency through MLflow's robust experiment tracking and model logging capabilities.
• Revolutionized deployment process, ensuring production-grade implementation and comprehensive documentation for replicability.

• Task 1: Design and implement a data pipeline using Apache Airflow to orchestrate ETL workflows, ensuring seamless integration and automated data processing across various data sources.
• Task 2: Set up real-time data streaming with Apache Kafka and manage distributed synchronization using Apache Zookeeper, facilitating robust, low-latency data ingestion and processing.
• Task 3: Configure and deploy a containerized data engineering environment with Docker, integrating Apache Spark for data processing and utilizing Cassandra and PostgreSQL for scalable and efficient data storage solutions.

• Exploratory Data Analysis: Understand the dataset structure, check for missing values, perform univariate and bivariate analysis, and summarize findings with visualizations.
• Feature Engineering & Machine Learning: Create new features, evaluate correlations, and select the best machine learning model based on evaluation metrics, providing insights into factors affecting timely deliveries.
• Application Development: Develop a Python Flask web application for users to upload data, perform analysis, and deliver insights, model results, and business recommendations.

• Analyzed athlete participation and medal counts across Olympic editions, identifying trends in sports and disciplines and highlighting athlete dominance in specific events.
• Provided a comprehensive overview of Olympic history and evolution, offering insights into athlete performance and sport-level trends.
• Utilized Azure Storage Containers, Databricks, Synapse Analytics, and Power BI for efficient data transformation and visualization, with findings aimed at informing future Olympic planning and inspiring athletes and fans.

• Task 1: Data Ingestion & Storage: Utilize AWS Glue to create a mechanism for ingesting data from multiple sources into Amazon S3, ensuring efficient and secure data transfer. Store the ingested data in a centralized data lake within S3 for unified storage and easy access.
• Task 2: ETL & Scalability: Use AWS Glue to transform raw data into the required format, and ensure scalability by leveraging AWS Lambda for scalable compute resources. As data volume grows, AWS Glue and Lambda will seamlessly handle the increased load without manual intervention.
• Task 3: Cloud Integration & Reporting: Manage access and security using AWS IAM. Use Amazon Athena to perform interactive queries on data stored in S3. Build a reporting dashboard with Amazon QuickSight to visualize data insights and answer key business questions, leveraging the cloud's scalability and performance.

• Prepared and cleaned data on top cryptocurrencies (e.g. Bitcoin, Ethereum, Binance Coin, Cardano) using R language.
• Analyzed cryptocurrencies for price trends, market share, and volatility using technical analysis (e.g. trend analysis and charting) and sentiment analysis (news articles).
• Conducted fundamental analysis to identify key factors (e.g. adoption rates, regulations, market sentiment) affecting cryptocurrency prices and used analytical techniques to develop a portfolio and propose risk management strategies for investing in volatile markets.

• Implemented two pipeline in AWS to categorise the Article/Conversation into 6 Classes (happy, angry, surprise, sad, disgust, not-relevant).
• Performed a comparative analysis of deep learning models including BERT, Roberta, and RNN to classify Twitter data into 6 emotions.
• Achieved 97.83% accuracy and 0.83 F1-score on RNN model and deployed the final model on Amazon SageMaker and Spark-MLlib.

• Performed Data Wrangling, Data preprocessing, Visualization and Analysis to predict price using Regression pipeline.

• Conducted data mining for 200000 tracks extracted by Spotify API, in order to analyze the trend of music industry development, and produce a predictive model for track popularity.

• Task 1: Data preparation and Customer Analytics: Collect, clean, and analyze customer data to gain insights about behavior and preferences.
• Task 2: Experimentation and Uplift Testing: Conducted controlled experiments and A/B testing to measure the impact of strategies and optimize business outcomes.
• Task 3: Analytics and Commercial Application: Applied data-driven insights to make informed decisions, formulate strategies, and drive commercial success.

• A Data Science project to analyze the customer feedback on quality service and improve the pitfalls.
• Utilize web scraping techniques to gain valuable insights about companies and predict customer buying behavior.

• Uncover actionable insights and visually communicate complex data patterns through interactive dashboards, empowering businesses to make data-driven decisions with confidence.

• Performed data analytics on Uber data using various tools and technologies, including GCP Storage, Python, Compute Instance, Mage Data Pipeline Tool, BigQuery, and Looker Studio.

• To design and develop a highly accurate real-time classification model capable of recognizing and interpreting American Sign Language (ASL) gestures from video data.

• Utilization of AI-guided algorithms for Covid-19 analysis of CT scans.
• Collection of 1,810 CT scan datasets, including 1,267 Covid-19 patients' and 543 healthy patients' scans.
• Training and evaluation of pre-trained models (InceptionNet V3 and U-net) using k-fold cross-validation to identify the most effective model for detecting, localizing, and segmenting Covid-19 cases in CT scan images.

• In this project, we collected Aerial Imagery datasets from OVERHEAD and MUNICH, analyze using python to solve Vehicle detection or sector related problems.
• Images were segmented using Region based method, HOG & SURF feature Extraction alogrithm were compared followed by SVM and Logistic regression.
• Statistical Results of Combinations of methods was reported. later we implmented Deep Learning Model (ACP-Fast CNN & Faster R-CNN). computed growth-truth, confusion-matrix, recall, precision, F1-score.

• Trained Neural Network with linear, sigmoid, and tanh activation functions.
• Utilized k-fold cross-validation (k=5) to evaluate performance.
• Computed confusion matrix, precision, recall, and F1-score to compare activation functions and determine the best-performing one.

This project was built for Academic Evaluation of term work in third year mini Project course of E&TC. we used 144 Leds, 6 4017-decade counter, 3 Shift register, resistors and 1 Long PCB to build this system. Using Andriod App connected via bluetooth module, custom text was shared and displayed on the board.

In this project, developed an Embedded System using miniature piston-solonoid to read and write for blind students. when the kid wish to read, he was able to read so direclty by attaching USB to system, when he wish to write, he could press solonoid pistons in pattern on Braille, the system was able to store it in USB file. The system was built through Arduino, Solonoid Piston(5mm), MySQL and provides the flexibility to add, modify or recreate new text-material for students.

Built Automated Bottle filling Plant System completely from scratch using Arduino, IR-sensors,dual directional-pump, bluetooth module, LCD display, transformer, potentiometer.

Designed a home automation tool, which can contol many smart devices like lights, mocks and even media devices. built upon Raspberry-Pi, controlled via Android Mobile App to display temperature and humidity. graphical display of temperature and humidity was also done using ThinkSpeak.

This is a tkinter based application where the list of books can be saved in the database. People can log in and can read books and can also hire or purchase books. The application is built through Python & MySQL.