Yashasvi Kotra

Yashasvi Kotra

Data Scientist
USA
Full-time On-site

About Me

Data Scientist with an M.S. in Applied Data Science and 3 years of experience building machine learning models in Python and SQL, from automated data collection and exploratory data analysis to deployment on AWS and Azur…

Experience

Data Analyst

Cargill · Remote
May 2025 - Oct 2026 · 1 year 5 months

<p>Automated data collection from 6 sources with scheduled Python scripts on AWS and cleaned 1.6M+ structured records and 3,000+ unstructured invoices with pandas and OCR preprocessing, cutting weekly prep for 6 analysts from 12 hours to 2.</p><p> Uncovered 3 recurring margin patterns through exploratory data analysis in Seaborn and Plotly, then presented them in Tableau and Excel to 15+ stakeholders, guiding $120K in purchasing.</p><p> Cut manual risk review 30% by training Random Forest and Decision Tree classifiers in scikit-learn (precision above 0.85 on 5-fold cross-validation), flagging about 40 bad records a month for 4 analysts.</p><p> Improved weekly demand forecast accuracy 18% (MAPE) by comparing Linear Regression and XGBoost models on 24 months of history, giving 6 finance leads a tested number before each purchase.</p><p> Classified 4,000+ invoice dispute notes into 6 reasons with NLP, comparing Naive Bayes and Support Vector Machine models in scikit-learn to reach 88% accuracy and save analysts 4 hours a week.</p><p> Reached 94% accuracy sorting 2,500 scanned invoice and delivery images into 5 document types with a TensorFlow convolutional neural network, removing 5 hours a week of manual sorting before OCR.</p><p> Deployed 4 machine learning models on AWS SageMaker with monthly retraining, holding precision above 0.85 for 6 months and cutting release time from 2 hours to 15 minutes.</p>

Data Analyst Intern

Symetra Investment Management · Connecticut
May 2024 - Aug 2024 · 3 months

<p>Raised ESG forecast accuracy 25% for 10+ portfolio analysts by training Linear Regression and XGBoost models on 3 years of market data, validated on 4 held-out quarters.</p><p> Extracted sentiment signals from 5,000+ financial news articles with NLP in Python, cutting forecast error a further 8%.</p><p> Streamlined ESG data processing 40% by consolidating 5 vendor feeds with Python and pandas, and automated the compliance Excel report from 5 hours to 3 per cycle.</p>

Data Engineer

Tata Consultancy Services, client Ecolab · Mumbai, India
Sep 2021 - Aug 2023 · 1 year 11 months

<p>Lowered demand forecast error (MAPE) 18% by training Linear Regression and XGBoost models on 3 years of order history, reviewed weekly with 4 data scientists.</p><p> Segmented 50K+ customers into 6 groups with K-Means clustering and RFM in Python, giving sales a ranked list of 2,000 accounts to contact first.</p><p> Moved model training and daily scoring to Azure Machine Learning with Azure Blob Storage inputs, so planners at 40+ sites had daily numbers.</p><p> Launched a Tableau view of on-time delivery and fill rate for 40+ sites, adopted by 6 regional managers within a month.</p>

Certifications

Google Professional Data Analytics Certification

Google · College Station , Texas , United States of America (USA) · 2025

Skills

Python NumPy pandas SciPy scikit-learn SQL Random Forest Decision Tree Linear Regression Logistic Regression Naive Bayes Support Vector Machine K-Means clustering XGBoost NLP forecasting anomaly detection TensorFlow convolutional neural networks CNN image classification PyTorch exploratory data analysis EDA data preprocessing structured data unstructured data Seaborn Plotly Matplotlib Tableau Excel Microsoft Azure Azure Machine Learning Blob Storage Amazon Web Services SageMaker S3 Git Jupyter
Report this Profile?