Keval Sakhiya

Skills

• Programming Languages: Python, SQL, JavaScript
• Data Science: Machine Learning, Deep Learning, Natural Language Processing (NLP), Data Analysis
• Frameworks & Tools: Pandas, NumPy, Matplotlib, TensorFlow, Keras, Scikit-learn, XGBoost, Power BI, etc.
• Databases: SQL, NoSQL
• Data Collection & Web Scraping: Scrapy, Requests, Selenium
• Data Engineering: ETL, Data Pipelines, Data Warehousing
• Cloud Platforms: AWS (EC2, S3), Azure, Google Cloud Platform (GCP)
• Version Control & Collaboration: Git, Docker
• Other Tools: MLflow, DVC, FastAPI, Django

Work Experience

Freelancer | Upwork

Data Scientist | 2022 - Present (2 years)

• Designed and implemented machine learning and deep learning models for diverse projects, delivering actionable insights and predictions.
• Applied MLOps best practices by integrating data version control and Git, ensuring the reproducibility and scalability of models.
• Implemented experiment tracking using MLflow and set up CI/CD pipelines with Git Actions to streamline model deployment.
• Tools used: Scikit-learn, TensorFlow, Keras, XGBoost, Matplotlib, Fast API, Power BI.

Data Engineer | 2019 - 2023 (4 years)

• Worked on numerous data gathering and web scraping projects, efficiently collecting and processing large datasets.
• Developed robust data pipelines to clean, transform, and store data in both SQL and NoSQL databases.
• Automated pipeline execution and scheduling for web scraping spiders, ensuring timely and accurate data retrieval.
• Collaborated with clients to design scalable solutions, utilizing cloud platforms such as AWS EC2 and S3.
• Tools used: Scrapy, Requests, Selenium, Git, Docker, SQL, NoSQL databases.

Projects

Emotion Detection Using CNN

GitHub: Emotion Detection Using CNN

Developed a deep learning model using a Convolutional Neural Network (CNN) based on the ResNet architecture to detect and classify human emotions from images. Achieved high accuracy through extensive data preprocessing and model tuning. Created a user interface with Gradio App for easy interaction and implemented an OpenCV-based app for real-time emotion detection via live video.

Property Scout - Real Estate Price Prediction

GitHub: Property Scout

Created a comprehensive real estate property price prediction system using XGBoost. The project includes a property recommendation system leveraging cosine similarity for location, facilities, and price, with a focus on MLOps practices including data version control and experiment tracking. This project helps potential homebuyers and real estate investors make informed decisions by accurately predicting property prices and suggesting optimal properties based on personalized preferences.

Movie Recommender System

GitHub: Movie Recommender System

Implemented a content-based movie recommender system that suggests movies to users based on their preferences. Utilized natural language processing (NLP) techniques to analyze movie metadata and calculate similarities using cosine distance, providing personalized recommendations.

Interests

• Blog Writing: Write and share data science insights on Medium, simplifying complex topics for a broad audience.
• AI Exploration: Learn and experiment with new AI technologies, including AI agents and APIs.
• Reading: Enjoy reading novels and books on technology, AI, and data science.