Reshma Am Mineni

Developed Grok-powered generative AI applications using Python, LangChain, and LangGraph for conversational AI, knowledge discovery, and task assistance. Built RAG knowledge systems using Pinecone and enterprise sources (PDF/Word/Excel/SharePoint, structured data, and web content) with OCR, semantic chunking, metadata enrichment, and embeddings. Engineered agentic workflows with tool calling, memory, task planning, and multi-step orchestration to handle complex user requests. Optimized response quality with prompt engineering, hybrid search, embedding and context/retrieval tuning to reduce hallucinations. Implemented LLM evaluation and production observability covering response quality, retrieval accuracy, latency, token usage, and reliability. Built scalable inference and integration services using FastAPI, Docker, Kubernetes, and AKS, including function calling to connect Grok with external APIs and enterprise systems while addressing reliability, security, and AI governance.

Engineered scalable machine learning and AI solutions on multi-million record datasets, improving accuracy by 20% while reducing processing time by 35% through optimized feature engineering and data pipelines. Created reusable ML frameworks for feature engineering, training, experimentation, and inference to standardize performance and speed up production deployments. Implemented transformer-based NLP solutions (BERT/RoBERTa via Hugging Face) for document classification, semantic search, entity extraction, and text analytics. Delivered production-grade inference APIs with FastAPI, Docker, Kubernetes, and MLflow for low-latency enterprise service deployment. Automated end-to-end MLOps using Airflow, MLflow, GitHub Actions, and CI/CD; added monitoring for drift, data quality, inference latency, and production performance to enable proactive improvements. Partnered with product and engineering teams to align model trade-offs and architectures with business goals.

Built end-to-end ML pipelines including ingestion, preprocessing, feature engineering, training, validation, and deployment. Developed predictive and NLP models using Scikit-learn, XGBoost, TensorFlow, and Hugging Face Transformers for customer analytics, forecasting, and intelligent document processing. Implemented ETL/data transformation pipelines using PySpark, SQL, and Apache Airflow. Created RESTful APIs using Flask/FastAPI to integrate ML models into enterprise business applications. Added model explainability with SHAP and feature importance analysis to improve transparency. Supported cloud deployment (AWS and Azure), along with monitoring and performance optimization, in Agile delivery teams.