Hire Jaswanth V. - Generative AI Engineer

At Wells Fargo, I worked as a generative AI engineer designing and deploying an internal, enterprise-grade multi-LLM chatbot to enhance document understanding and retrieval across financial services workflows. I integrated AWS Textract OCR to improve extraction quality from scanned and semi-structured PDFs and built RAG pipelines using LangChain with FAISS and Pinecone to enable accurate, context-aware querying while reducing hallucinations and improving response fidelity. I developed prompt templates and schema-aware validation mechanisms to ensure data accuracy, robustness, and compliance with internal standards. The system was deployed using FastAPI and Docker with CI/CD pipelines on AWS and instrumented with MLflow, Weights & Biases, and Grafana for monitoring, evaluation, and performance tracking. This work accelerated access to regulatory and operational data, enabling faster, more informed decision-making in complex, regulated financial environments.

At Honeywell, I worked as a machine learning engineer converting regulatory PDFs into structured data via Python parsing tools and XML/CSV pipelines for advanced analytics and regulatory compliance. I developed robust NLP preprocessing pipelines using spaCy and NLTK to normalize noisy text and tables, enabling consistent and accurate model inputs. Built and deployed sentiment and spam classification models to enhance data quality and compliance alert mechanisms, implementing schema validation and automated document tagging. I deployed workflows using FastAPI, Docker, and orchestration tools like Apache Airflow and Prefect, while integrating MLflow for experiment tracking and Streamlit dashboards for real-time monitoring and stakeholder reporting. My contributions accelerated data extraction efficiencies and improved automated content classification, supporting Honeywell’s compliance and risk management initiatives.

* At Accenture, I contributed as a data scientist, focusing on demand forecasting, estimated time of arrival (ETA), and route efficiency optimization for logistics operations. I developed and fine-tuned machine learning models using Python, scikit-learn, and XGBoost to improve the accuracy and reliability of delivery planning and routing predictions. I managed large-scale data processing workflows with Apache Spark and Hadoop, automating data ingestion and pipeline orchestration using Apache Oozie. I implemented model training and deployment pipelines on CUDA-enabled AWS infrastructure to ensure scalable and efficient prediction serving. My work included integrating predictions with Redshift and NoSQL databases to enable low-latency access by downstream applications. Additionally, I performed extensive model validation, monitored for seasonal data drift, and implemented mechanisms for continuous performance evaluation, supporting operational decision-making and business growth.