Senior AI/ML Engineer at Stripe

Built production-grade ML and Generative AI systems for payments intelligence and fraud investigation. Designed and deployed real-time fraud detection and payment risk scoring models with Python, XGBoost, and PyTorch, and served them via FastAPI microservices with sub-150ms latency. Implemented RAG and hybrid retrieval for fraud investigation workflows using LangChain/LangGraph and embeddings with reranking. Developed agentic, multi-agent fraud investigation orchestration with tool/function calling and human-in-the-loop validation using LangGraph. Added LLM guardrails (output validation, hallucination detection, prompt-injection protection) and implemented evaluation pipelines with RAGAS/DeepEval/TruLens for retrieval and response quality. Ensured production reliability using MLOps, monitoring, observability (LangSmith/MLflow/Prometheus/Grafana), and secure integrations via Model Context Protocol.

ML Engineer at Dollar General

Developed scalable demand forecasting and inventory optimization solutions for retail operations across thousands of stores. Built ML and forecasting models (ARIMA, Prophet) and price/promotion analysis (elasticity and impact). Created an LLM-powered demand forecasting copilot and a conversational analytics application using LangChain, RAG, and vector retrieval. Implemented LLM-assisted text-to-SQL with schema grounding and access controls. Built data pipelines using Databricks, Delta Lake, PySpark, and MLflow for repeatable training and deployment, and implemented model monitoring and explainability using SHAP. Developed REST APIs for analytics and retrieval services and supported A/B testing and statistical analysis for pricing and marketing.

Data Engineer – Conversational AI Platforms at Aisera

Built and maintained conversational data pipelines and enterprise integration systems to support AI-driven IT service automation. Prepared large-scale conversational datasets for NLP models (intent classification, entity recognition) and developed processing pipelines for chatbot logs and user interactions. Integrated conversational workflows with ServiceNow, Salesforce, Jira, and enterprise databases via REST APIs for ticket routing and knowledge access. Implemented ingestion, chunking, indexing, and retrieval pipelines for structured and unstructured knowledge sources. Supported prompt design, testing, fallback handling, and continuous improvement using interaction logs. Deployed models via FastAPI and ensured data quality and reliability for training and inference.