I’m a data enthusiast who love exploring data sets related to all fields of science and technology.
Using a Stack Overflow dataset on questions and answers on python programming language. The objective is to build a chatbot for answering user questions based on python programming language.
• Analyze and preprocess the data to understand patterns in questions and answers.
• Build intents to enable the chatbot identify predefined patterns and responses.
• Design a deep learning model to train a chatbot for answering python based user questions.
Github Link: Coming Soon
Jupyter Notebook HTML Link: Coming Soon
Collected fake & real news dataset from Kaggle. I found that interesting as fake news is one of the leading topics of discussion in the world.
• Perform data analysis using a jupyter notebook to identify fake/real news structure.
• Conduct n-gram analysis and topic modeling, to understand news topics and patterns.
• Implement data cleaning to remove bad symbols, stop words and lemmatization.
• Build global vector word embeddings to build model features.
• Incorporate a deep learning model to classify fake or real news with 96% accuracy.
• Apply chi2 statistical analysis to identify words impacting fake and real news.
Github Link: Link
Jupyter Notebook HTML Link: Link
Worked with Intuit for more than 2 years on the intersection of data scientist and machine learning engineer.
Even though I was a part of security and fraud team, I got a chance to work on a multitude of problems realted to products.
I explored diverse datasets related to product/application users, security/network, text and time-series.
Responsibilities:
• Perform data analysis on varied datasets, to discover opportunities in product security improvement and adversary management.
• Build deep learning models to forecast overall product traffic (users and adversaries). Alerting on product’s normal and anomalous user behaviors.
• Design and implement machine learning model to classify CVE's based on text description, for improving remediation process.
• Collaborate with data engineering team to build large scale data ingestion pipelines in AWS, using AWS Glue and AWS EKS (Spark on EKS).
• Refined recommendations and experimentation results data, utilised by product and security teams for product enhancements.
Started working with Starva as part of their thesis/capstone project. Strava is a website and mobile app used to track athletic activity via satellite navigation. Its is also called "The Social Network for Athletes”.
Responsibilities:
• Analyze geospatial data covering cycling patterns and trends of Strava mobile application users in San Francisco area.
• Build deep learning forecasting model to predict the increase/decrease in Strava userbase.
• Perform exploratory and heat map analysis to analyze trend between increasing bike crimes and Strava user activities.
Started working with a start-up to gain exposure on their operations and challenges. It was an enriching experience, gave me an opportunity to lead a team and set up data science processes from scratch.
Responsibilities:
• Mentor a team to catalog data sources, identify gaps in data capture and build processes to maintain data hygiene.
• Build data ingestion pipelines for product team using AWS Kinesis and Lambda.
• Design a sale improvement prototype, for household products recommendation engine using customer transaction data.
Acompworld combines agile and design thinking practices to deliver digital platforms.
Responsibilities:
• Optimize the sales process through product sales and customer data analysis, it increased 20% of the annual sales.
• Assist the marketing head in developing company’s new market in 5 different states.
• Generate ad-hoc reports on purchased products and track changes in overall customer base.
• Coordinate with the engineering team to develop new product prototypes using design thinking.