Shivansh Shrivastava

I’m a Data Scientist focused on designing, developing, and implementing AI/ML models that move from training and evaluation to deployment. I work with Python-driven pipelines to deliver AI-driven solutions with measurable accuracy improvements.

In my Data Science & Analytics Intern role, I collected, cleaned, and validated data across 100K+ records, using SQL joins, aggregations, and CTEs to improve data quality—reducing inconsistencies by 30% and raising downstream validation scores from 81% to 89%. I also automated analysis with Python and delivered dashboards and reporting with Power BI, Matplotlib, and Seaborn.

I bring strong analytical depth through statistical analysis like regression analysis, hypothesis testing, clustering, and time-series thinking—turning business datasets into actionable insights for cross-functional stakeholders with 88%+ analytical accuracy. I’m comfortable iterating fast, documenting clearly, and translating technical findings into decisions.

My projects reflect end-to-end ownership: I built an LSTM-based forest fire risk predictor (Pyro Predict) with an F1 score of 0.67 and recall improvements (0.79 to 0.89), and an NLP-driven resume-JD compatibility scorer (Resume Lens) reaching 85%+ semantic accuracy while cutting resume-to-JD comparison time by 95%. I also implemented and deployed an inference solution (Ship Sense) using FastAPI and Docker, achieving an average warm inference latency of 351ms and reducing RMSE by 18% using feature engineering and GridSearchCV tuning.