Senior Data Scientist – AIOps at Jefferies Group

Developed anomaly detection models for infrastructure logs, system metrics, and telemetry to support proactive incident identification. Built RAG pipelines (LangChain) indexing incident tickets, operational runbooks, and change records for context retrieval during troubleshooting. Implemented online-offline feature workflows and engineered time-series datasets with missing/duplicate handling, irregular sampling fixes, and rolling-window/seasonal/statistical feature generation. Designed agentic AI workflows (LangChain/LangGraph) for multi-step incident investigation using enterprise tools/APIs with workflow state orchestration. Applied SHAP/LIME for interpretability, investigated false positives/negatives for RCAs, and established automated LLM/agent evaluation pipelines for retrieval relevance, response quality, and tool-selection accuracy. Supported secure ML operations by integrating Azure Key Vault/HashiCorp Vault for secrets; managed experiments and model lifecycle via MLflow and A

Data Scientist – MLOps at Comerica Bank

Engineered risk models by creating behavioral/transactional/repayment features from Snowflake datasets and supporting risk assessment workflows. Built repeatable data preparation and deployment processes using Databricks and Azure Blob Storage for raw datasets, intermediate features, and artifacts. Extracted structured entities from financial documents using Hugging Face transformer models for downstream credit risk and regulatory reporting. Developed and deployed REST inference APIs using Flask and Docker; applied SHAP/LIME for explainability and compliance support. Addressed class imbalance for fraud detection and built dbt models for analytical transformations. Implemented Feast feature management to define reusable feature views and validated offline/online consistency. Created automated data validation to ensure schema/distribution correctness before training; managed experiments and model lifecycle with MLflow; scheduled and supported governance-ready releases.

Data Scientist – NLP at Quantiphi

Developed NER solutions by defining annotation guidelines, building labeled corpora, evaluating entity extraction, analyzing errors, and iteratively improving annotation quality. Built Flask-based REST APIs for NLP inference with input validation, exception handling, logging, and integration to internal apps. Created text classification models (scikit-learn, XGBoost) for support request categorization; tuned hyperparameters for better generalization across datasets. Containerized inference services with Docker for consistent deployment across environments. Performed robust data profiling/cleansing, standardization, and feature engineering for classification/NER/sentiment; migrated R scripts to Python while preserving analytical consistency. Loaded processed datasets and inference results into databases with schema validation; collaborated with BAs to translate reporting/document transformation rules into analytics workflows.