Tejaswini Matti

• • Extracted and preprocessed 10,000+ ECS Wiki articles (HTML, Plain Text, Markdown) from a HANA vector database using SpaCy lemmatization for high-quality input data.

• • Built an end-to-end unsupervised NLP pipeline using BERTopic, generating 250+ coherent topics from 10,000+ articles via transformer-based embeddings and clustering.

• • Designed a Neo4j Knowledge Graph with 3 entity types and 10,000+ nodes, enriching metadata and surfacing contextual relationships across the wiki corpus.

• • Integrated a RAG chain retriever and developed a Streamlit frontend, enabling semantic search across 10,000+ articles and 250+ topic clusters for improved content discoverability.

• • Led API and backend testing for large-scale digital payment systems processing lakhs of transactions across 3+ banking partners and 4 APIs.

• • Designed UAT, regression, and functional test strategies; validated complex transaction workflows and compliance requirements over 1.5+ years.

• • Created 500+ test cases with a 75% pass rate, and validated weekly GEFU and Merchant Payment Reports (MPR), ensuring payment accuracy across distributed workflows.