Khyati Sharma

3390 Stratford RD NE · Atlanta, GA 30326 · (404) 390-5306 · sharmak9610@gmail.com

Buckle up and join me on a thrilling ride through the world of data science in the fast-paced automotive industry. With four years of navigating the twists and turns of supply chain and customer analytics, I've honed my skills to turbocharge business performance and accelerate growth.

As your Data Scientist guide, I bring more than just horsepower to the table. I've fine-tuned my engines to harness the raw power of machine learning models, neural networks, and language models, transforming data into actionable insights that drive decisions and steer strategies towards success.

From decoding demand fluctuations to fine-tuning inventory management, I've engineered solutions that rev up efficiency and revitalize customer experiences. But what really fuels my passion? It's the thrill of uncovering hidden patterns, the rush of optimizing operations, and the satisfaction of seeing businesses roar ahead in the race to stay ahead of the competition.

But what sets me apart from the rest? It's not just about building machine learning models or tweaking neural networks—it's about crafting bespoke solutions that revitalize businesses and fuel growth. Whether it's steering companies towards smarter decisions or turbocharging customer experiences, I'm committed to delivering results that leave a lasting impact.

Let's turbocharge the data journey and accelerate toward a future where insights drive innovation and success fuels ambition!

Experience

Georgia State University

Graduate Research Assistant

As a Graduate Research Assistant at Georgia State University, I had the privilege of contributing to the academic growth and development of students while gaining valuable experience in higher education. My role encompassed a wide range of responsibilities, including:

  • Course Facilitation: Assisting professors in leading and facilitating class discussions, and lab sessions, creating an engaging and interactive learning environment.
  • Research contribution: Contributing to research in NLP data mining process under professor.
August 2023 - Present

Tata Consultancy Services

Data Scientist

At TCS, I had the opportunity to work closely with Jaguar Land Rover as both a Data Engineer and Data Scientist. This dual role allowed me to leverage my expertise effectively, contributing directly to the success of Jaguar Land Rover while honing my skills in the automotive industry.

1. Optimized financial data processes by leading SAP analysis and implementing end-to-end ELT with Dataform:

  • Automated data extraction and transformation for real-time financial insights.
  • Established a cash flow data warehouse using Google Cloud Composer, resulting in a 30% reduction in processing time.
  • Collaborated cross-functionally to translate business requirements into scalable cloud-based data solutions, aligning with GCP best practices.
  • Received recognition from client (Finance department head for Jaguar Land Rover) for rapid development of tool.

2. Managed supply chain risk by creating a data warehouse, and storyboards in Tableau, utilizing Python for data extraction from Maplecroft API and SC database in GCP:

  • Engineered a Dataform-driven ETL pipeline, orchestrated through Apache Airflow for efficient data management.
  • Implemented unit testing in Python as a proactive measure to enhance code quality and streamline integration and deployment processes during productionization. This initiative resulted in improved code reliability and functionality, ensuring smoother transitions from development to production phases.
  • Created DAGs to orchestrate the pipeline flow, resulting in an 80% improvement in time efficiency and operational effectiveness.

3. SocialSentinel- Customer Sentiment Analysis for product success:

  • Applied NLP-based sentiment analysis algorithm, utilizing Random Forest, and Support Vector Machine algorithms to predict product success with 88% accuracy and asses product at feature level.
  • Developed web scraping model, achieving a data retrieval accuracy rate of 95% and reducing data collection time by 60%, thereby establishing the foundation for robust, data-driven analysis.
  • Utilized text cleaning and feature engineering. This project allowed the Stakeholders to take decisions if recalling of batch was required based on the feature analysis done from the customers behaviour.
  • Successfully implemented CI/CD pipelines using Gitlab for version control on GCP, reducing deployment time by 40% and ensuring reliable and consistent releases.
March 2022 - July 2023

Bristlecone Inc.

Data Scientist

As a data scientist at Bristlecone, I had opprtunity to work closely with Mahindra Groups and explore dynamics of Supply Chain, Logistics and Warranty analytics by implementing analytical solutions. I was responsible for follwoing projects:

1. Business Decision Support Systems Development:

  • Developed a Flask-based decision support system integrating CNN algorithm for image recognition of corrosive products, achieving 94% accuracy improvement in product defect detection.
  • This project encompassed from enabling Stakeholder to make decisions from the analytics report generated from this dasboard.
  • The Task time was reduced from 24 working days to 1 hour, resulting in $10+ million in savings.
  • Provided a user-friendly web interface and data access through RESTful APIs.
  • Received Innovation Award, an annual accolade reserved for a single employee who pioneers distinctive and impactful innovations.

2. Demand Forecasting and Strategy for Long-Term Capacity:

  • Developed R scripts to translate demand forecasts into manufacturing-compliant blueprints.
  • Created interactive R Shiny modules and utilized data analysis and visualization tools for insightful decision-making.

3. Route Optimization for Outbound Logistics:

  • Led the team and implemented route optimization model by leveraging OpenStreetMap data and HERE API to optimize routes.
  • Used statistical analysis, visualization, and cleaning techniques to help with the optimum route in the model.
  • Successfully achieved 50% improvement in average transit time by implementing advanced algorithm tailored for truck-type vehicles.

4. Finished Goods Inventory Optimization:

  • Performed ABC-XYZ analysis to identify low and high inventory hotspots.
  • Implemented Exponential Smoothing Time Series model to forecast the demand
  • Analysed data sources from manufacturing, sales, billing to build a central database using ETL process.
  • Created Tableau dashboard providing real-time insights into inventory levels, turnover rates, and product availability across multiple warehouses aiding business stakeholders in resolving stock-out situations, and making data-driven decisions.

5. Vehicle Off Roading Analysis:

  • Using python, analysed data with ABC classification methodology to identify frequent part at dealer and vehicle level and push dealer to always keep X units in stock
  • This helped Stakeholders to identify dealers which caused company to incurr huge loss due to not keeping the stocks and detect anomalies in the billing cycle
  • Recognized by client's CFO with awards and commendation for adept cost analysis, insights derivation, and anomaly detection.

6. Customer Value Analysis:

  • Implemented ETL-driven data extraction in the SAS Enterprise Guide to analyze customer value.
  • Developed a dashboard on SAS Visual Analytics to identify customer value, focusing on recency, frequency, and monetary analysis.

September 2019 - February 2022

Projects

Trading App for Investments | NLP, Prediction, Neural Networks, Web Scrapping, Flask, GCP, AWS

  • A complete end-to-end platform for stocks trading and utilizing stock price prediction module for decision making of stock buying or selling.
  • This application provides real time trading and portfolio management.
  • Developed a Dynamic Linear model to predict stocks low and high of the day.
  • Utilized BERT for news article sentiment analysis to provide insights into market movements.
  • Provided financial chatbot through open AI integration.
  • Developed an interactive web page on react js and backend in Flask.
  • Deployed the project on GCP and backend in AWS EC2.
  • Application Link: "https://wetrade-4b06f.web.app/"

Travel Recommendation System | Content-Based Filtering

  • Utilized logistic regression modeling to predict user interest in travel destinations based on historical data and destination attributes, enhancing our ability to anticipate user preferences and behaviors.
  • Employed the logistic regression predictions to adjust the feature vectors of destinations, weighting them according to the predicted likelihoods of user interest, thereby prioritizing destinations with higher anticipated relevance to the user.
  • Integrated cosine similarity calculations with the adjusted destination feature vectors and user preferences to determine the relevance of destinations to individual users, ensuring highly personalized recommendations aligned with user preferences and anticipated interests.

Text Classification Model | Advanced Natural Language Processing (NLP)

  • Conducted a comprehensive evaluation of text classification models utilizing advanced Natural Language Processing (NLP) techniques.
  • Explored a range of recurrent and convolutional neural network architectures, including RNN, LSTM, GRU, Bidirectional LSTM, Bidirectional GRU, and 1D Convnet models.
  • Leveraged the Reuters newswire dataset for training and evaluation.
  • Rigorously evaluated and compared the performance of each model.
  • Provided valuable insights into the effectiveness of different neural network architectures for text classification tasks.

Caption Generation using CNN | Advanced Natural Language Processing (NLP)

  • Developed an innovative system aimed at automatically generating descriptive captions for diverse image datasets, catering to applications such as image indexing, content retrieval, and accessibility enhancement for visually impaired individuals.
  • Employed Convolutional Neural Networks (CNN) in conjunction with deep learning and natural language processing techniques to realize the project's objectives effectively.
  • Incorporated essential callbacks such as ModelCheckpoint, EarlyStopping, and ReduceLROnPlateau to monitor and optimize the training process effectively. These callbacks ensured the model's stability, prevented overfitting, and facilitated efficient convergence towards optimal performance.
  • The utilization of CNN facilitated robust feature extraction from images, enabling the generation of accurate and contextually relevant captions.
  • This project represents a significant advancement in leveraging cutting-edge technology to enhance image understanding and accessibility across various domains.

Airline Service Sentiment Intelligence: Leveraging Enhanced Features for Customer Feedback Analysis | Advanced Natural Language Processing (NLP)

  • Developed a sentiment analysis system tailored for airline service reviews, aimed at predicting the success of the provided services.
  • Explored various machine learning models including Random Forest, Naive Bayes, and multinomial Logistic Regression for sentiment classification.
  • Utilized the NLTK library's VADER model for initial sentiment labeling based on compound scores.
  • Trained the machine learning models on labeled reviews to predict sentiments and understand the service performance.
  • This project represents actionable insights for improving airline service based on sentiment analysis of customer feedback.

Context-Aware Chatbot Development | Advanced Natural Language Processing (NLP)

  • Developed a specialized conversational chatbot tailored for academic inquiries, focusing on the academic domain's specific requirements.
  • Implemented LDA topic modeling for extarcting topics from question and refining our search and making it robust and efficient.
  • Used Cosine Similarity based vector search in MongoDB to find answer for the question from our DB which is closest match.
  • TF-IDF vectorizer for vectorizing the questions.
  • Created UI on Gradio

Education

Robinson College of Business - Georgia State University

Masters of Science
Computer Information Systems - Big Data Analytics and Management Track

GPA: 3.83

August 2023 - July 2024

Vidyalankar Institute of Technology

Bachelor of Engineering
Engineering - Electronics & Telecommunication Track

GPA: 3.2

June 2014 - June 2018

Certificates & Publications

Business Analytics and R

offered by Edvancer Eduventures

Certificate no.: 0101-130819-01

December 2018 - August 2019

Data Analytics using SQL

offered by Edvancer Eduventures

Certificate no.: 0601-220819-02

December 2018 - August 2019

Publications

Precise Angular Rotation Measurement with Sagnac Interferometer (published offline):

  • Presented and published at 26th DAE-BRNS National Laser Symposium held at Bhabha Atomic Research Center on December 2017.
  • resented and published at International Conference on Photonics and High-Speed Optical Networks 2018 in association with IEEE Photonics Society Madras Chapter.

Technical Poster Presentation:

  • Presented and published a poster on Precise Angular Rotation Measurement using Optical Fiber Gyroscope at the 3rd International Conference on Microwave and Photonics organized by IIT Dhanbad on February 2018.

Skills

Programming Languages & Tools
Workflow
  • Problem Definition & Data Collection
  • Data Cleaning & Preparation
  • Exploratory Data Analysis (EDA)
  • Feature Engineering & Selection
  • Model Selection & Developmen
  • Model Evaluation & Validation
  • Deployment & Integration
  • Monitoring & Maintenance
  • Collaboration & Communication
  • Agile Development & Scrum

Awards

  • Recognized by client's CFO with awards and commendation for adept cost analysis, insights derivation, and anomaly detection.
  • Received Innovation Award, an annual accolade reserved for a single employee who pioneers distinctive and impactful innovations.
  • Received Certification of Appreciation from client (Mahindra and Mahindra) for outstanding contribution in Rust – Text Analytics and Image Tagging Project.
  • Remote and Stellar Award from Bristlecone Inc. for contribution in project helped improved efficiency levels multiple folds through saved cost, time, and effort.
  • Brilliant Beginner Award for good performance and attitude within short span of time.
  • Spot Award for being proactive with quick problem-solving skills and ability to work in multiple projects at a time.
  • Xtra Miler Award For demonstrating outstanding performance above and beyond regular scope of work.
  • Team Excellence Award for recognition of outstanding teamwork and dedication to achieving exceptional results.
  • Kaizen Award for FY21 from Bristlecone Inc.