Vasavi Mekala
Ai Ml Engineer @Allstate
Signup · Get unlimited contacts
WORK HISTORY
Ai Ml Engineer @Allstate
Dallas, TX, US
Developed predictive models using LightGBM and XGBoost to identify high-risk accounts for payment defaults, using transaction histories, repayment patterns, and behavior data. Built a risk scoring matrix across accounts to guide collections and reduce exposure.•Implemented GPT-powered decision support with LangChain, integrating AI outputs with internal finance knowledge bases and policy rules. Delivered interpretable recommendations for credit approvals, improving turnaround time and ensuring consistency across product lines.•Engineered anomaly detection systems using Scikit-learn and AutoML frameworks to monitor over 1.1 million daily transactions. Used clustering and distance matrices to detect unusual patterns, enabling early fraud alerts and reducing escalations by 40 cases weekly.•Built an internal NLP chatbot using BERT and Hugging Face Transformers for employee support across finance and operations. Handled monthly queries and used a topic coverage matrix to ensure recurring issues were automated, freeing 15+ staff hours weekly.•Designed LSTM-based recommendation models using PyTorch and Scikit-learn for personalized insurance and installment offers. Developed a customer preference matrix to rank offers by predicted engagement, improving adoption for targeted campaigns.•Built automated ML pipelines for credit, fraud, and promotion models with Airflow/MLflow; tracked reliability and drift via dashboards, saving 30 hours/month; created Power BI performance dashboards; implemented reinforcement learning in TensorFlow/Keras to optimize credit line approvals.
EDUCATION
University of North Texas
Master of Science - MS, Computer Science
SJB Institute of Technology
Bachelor of Engineering - BE, Computer Science
ABOUT VASAVI MEKALA
AI/ML Engineer & Data Scientist with 4+ years of experience delivering predictive, NLP, and recommendation solutions in financial services and retail. Skilled in Python, PyTorch, TensorFlow, Keras, Scikit-learn, Pandas, NumPy, SQL, PySpark, and LSTM networks, with expertise in developing models for credit risk, churn, pricing, and fraud detection. Experienced with NLP and Generative AI using GPT-4, BERT, Hugging Face Transformers, LangChain, and LLaMA for document classification, semantic search, and AI-powered decision support. Built and deployed interactive Power BI dashboards integrating outputs from predictive, recommendation, and anomaly detection models, leveraging matrices such as churn scoring, product affinity, price elasticity, alert, comparative performance, and decision impact to drive actionable business outcomes across transactions, 1M+ daily payments, monthly inquiries, and accounts.
This profile is compiled from publicly available professional sources. Unifers is not affiliated with or endorsed by LinkedIn. Request removal of this profile.