Veltris - Senior Data Scientist - Agentic AI Platform

Veltris

Hyderabad

5 - 7 Years

Data Science Data Scientist Agentic AI Artificial Intelligence NLP Machine Learning RAG Python PyTorch Tensorflow

Posted on: 13/03/2026

Job Description

Senior Data Scientist- Agentic AI Platform

Experience : 5+ years in applied data science, with 3+ years in financial domain AI

Role Overview :

We are hiring a Senior Data Scientist to develop and deploy advanced AI/ML models powering mortgage automation agents. This role involves designing models for creditworthiness prediction, fraud detection, income verification, document intelligence, and conversational agents. The candidate will ensure models are accurate, explainable, and compliant with regulatory frameworks.

Key Responsibilities :

- Build, train, and optimize ML/AI models for mortgage risk, loan eligibility, pricing, and fraud detection.

- Develop document understanding and NLP models for parsing tax returns, pay stubs, W2s, bank statements.

- Implement RAG pipelines for domain-specific knowledge retrieval.

- Ensure bias mitigation, model fairness, and explainability (XAI) for compliance with regulatory audits.

- Collaborate with AI architects to integrate models into the agent framework.

- Develop monitoring dashboards for model drift, accuracy, and performance.

- Work with mortgage SMEs to translate domain knowledge into features, embeddings, and rules.

Required Skills & Experience :

- Strong experience with Python, PyTorch, TensorFlow, Scikit-learn, Hugging Face Transformers.

- NLP expertise : NER, summarization, Q&A, OCR (Tesseract, Amazon Textract, Azure Form Recognizer).

- Deep understanding of statistical modeling, risk scoring, credit analytics, fraud detection.

- Experience with vector databases (Pinecone, Weaviate, FAISS, Milvus).

- Knowledge of MLOps tools (MLflow, Kubeflow, Weights & Biases).

- Familiarity with mortgage underwriting, risk modeling, KYC/AML checks.

- Strong foundation in responsible AI practices (fairness, accountability, transparency)