Tejaswini Waghmare

About

Worked on large financial analytics datasets combining complaint text, transaction logs, and demographics; built end-to-end NLP/ML pipelines (TF-IDF + Random Forest) and data integration in BigQuery with Tableau reporting, citing ~95–98% accuracy. Also implemented entity resolution with fuzzy matching and semantic linking using BERT sentence-transformer embeddings stored in FAISS, including fine-tuning on labeled pairs to improve search/linking relevance.

Experience

Education

Key Strengths

• Unifying structured and unstructured financial data into a single analytical layer
• Building NLP classification pipelines (TF-IDF + classical ML) with strong reported accuracy (~95–98%)
• Entity resolution using hybrid exact + fuzzy matching with precision/recall validation
• Semantic search and record linking using BERT-based embeddings with FAISS vector database
• Fine-tuning embedding models on labeled pairs to improve relevance and reduce false matches
• Designing production-grade Python workflows with modularity, logging, error handling, orchestration, and tests
• Pragmatic transition from experimentation to production (stability/requirements-driven)