As a Principal Data Scientist with over 7 years in finance, I design and deploy end-to-end AI and ML solutions covering credit risk underwriting, recommendation systems, forecasting, fraud detection, data platforms, big data, and MLOps. I build scalable, fair, and explainable models aligned with business goals and regulatory standards, enabling financial institutions to make responsible, data-driven decisions. I led the development of Foneloan, Nepal’s first AI-powered instant lending platform, managing technical feasibility, data integration, model design, and deployment. This core credit scoring engine supports multiple A-level banks and manages billions in loan portfolios. Beyond Foneloan, I have developed and operationalized diverse models and platforms across retail and SME lending, plus other critical financial services. ✅ Key Focus Areas - Aligning AI and ML strategy with underwriting and business objectives - Translating credit needs into data science solutions - Breaking down credit risk projects and optimizing team resources - Leading collaboration across data, product, engineering, and compliance - Innovating with behavioral, transactional, and alternative data features - Leading research on credit scoring, explainability, fairness, and rapid prototyping - Designing scalable credit risk architectures with pipelines for ingestion, modeling, and monitoring - Managing model lifecycle including development, validation, calibration, and documentation - Enforcing governance through standards, versioning, monitoring, and audit readiness - Promoting responsible AI with fairness, transparency, and explainability - Mentoring data scientists and engaging with regulators and stakeholders ✅ Responsibilities - Built risk pipelines for origination, management, and collections - Led PoCs and feasibility studies for new credit data initiatives - Researched alternative data to improve risk segmentation - Managed model governance, documentation, and audits - Structured data science workflows and validations - Presented models to regulators, auditors, and executives ✅ Stack - Languages: Python, R - MLOps: MLflow, Feast, DVC - Databases: MySQL, Oracle, MSSQL, SQLite, MongoDB - Big Data: Hadoop, Spark, Sqoop, HBase, Hive, Kafka, NiFi, Druid, Kylin - Visualization: PowerBI, Tableau, ClickView, Superset, Plotly, Seaborn, Matplotlib - Job Scheduling: Crontab, Airflow - ETL Tools: Pentaho, pandas, PySpark, dbt - Cloud: AWS (EC2, EMR, S3, Lambda) - Containers: Docker, Kubernetes - Project Mgmt: Git, Jira, Trello - Dev Methodologies: Scrum, Kanban