Shivam Choudhary

WORK HISTORY

EDUCATION

ABOUT SHIVAM CHOUDHARY

Data Scientist & Machine Learning Engineer with 5+ years of experience delivering AI, machine learning, and data analytics solutions across finance, manufacturing, and regulatory domains. Proficient in Python, SQL, PyTorch, Scikit-learn, NLP, time series forecasting, deep learning, cloud (AWS, GCP, Azure), MLOps, and big data. Strong track record of designing, developing, and deploying models that improve accuracy, reduce costs, and drive business impact- Honeywell (AI/ML Intern) – Designed and implemented triplet mining strategies (hard/hybrid) for asset fault classification embedding models, boosting silhouette score from 0.72 to 0.9 and improving classification accuracy. Refactored DBSCAN clustering pipeline to run directly from Databricks tables, adding unit tests for scalability and robustness- American Express / Accertify (Senior Analyst, Data Science) – Built and iterated XGBoost and GBM fraud detection models for 30+ enterprise clients, achieving $14M+ monthly savings, 80% fraud coverage improvement, and up to 30% review rate reduction. Delivered targeted fraud strategies for retail and airline clients, cutting approved fraud dollars by up to 20%- Reserve Bank Information Technology (Data Scientist) – Developed BERT-based question answering systems and implemented unsupervised learning (DBSCAN, k-means, DTW) to analyze borrower default patterns for RBI’s statistics department, improving risk insights- Quantiphi (Machine Learning Engineer) – Built and deployed time series forecasting models (ARIMA, ETS, LSTM, Prophet, XGBoost) with stacked meta-learners, improving forecast accuracy by 20% and cutting runtime from 20 minutes/store to 1 minute/store using vectorization. Delivered deep learning NLP solutions (LSTM + GloVe) for genre classification and optimized ML pipelines for production environments.Passionate about end-to-end ML solution development, from data preprocessing and model training to deployment and monitoring. Thrive on applying time series analysis, NLP, anomaly detection, and cloud-based ML pipelines to solve challenging business problems.Want to connect? Feel free to email me at s••••••••@gmail.com or drop a hello on LinkedIn.