AI/ML Data Scientist

Qtsolv; Bengaluru; On-site; INR 4,000,000 - 7,000,000; Full time

Job summary

Qtsolv is seeking an experienced AI/ML Data Scientist to design, build, and deploy ML models across our payments platform, from real-time fraud detection to authorization optimization and risk scoring. You’ll work with Fraud, Risk, Engineering, Compliance, and Product to productionize models that run in high-volume, low-latency environments.

This role sits at the intersection of data science, payments domain expertise, and production ML engineering, with a focus on explainability, monitoring,

Qualifications

• 10+ years of overall experience in data science or ML engineering, with 3+ years in production environments.
• Strong proficiency in Python (pandas, scikit-learn, XGBoost/LightGBM, PyTorch or TensorFlow) and SQL.
• Experience building and deploying models into production with monitoring and retraining considerations.

Responsibilities

• Design, train, validate, and deploy ML models for real-time fraud detection and risk scoring.
• Engineer features from transactional and behavioral data to meet sub-100ms latency at decision time.
• Develop model monitoring for drift and data quality with KPI-based alerting.
• Establish and follow MLOps practices: versioning, reproducible pipelines, CI/CD for models, and testing frameworks.

Skills

Python

SQL

Pandas

ML Deployment

ML Frameworks

MLOps

Communication

Education

MSc/PhD in Statistics, CS, OR

Tools

AWS

GCP

Azure

SageMaker

Vertex AI

Databricks

Spark

Airflow

Job description

We are looking for an AI/ML Data Scientistto design, build, and deploy machine learning models across our payments platform - from real-time transaction fraud detection to authorization optimization, credit/behavioral risk scoring, and anomaly detection in scheme and settlement data. You'll work with high-volume, high-velocity transactional data in a regulated, low-latency environment where model accuracy, explainability, and operational robustness all matter as much as raw predictive performance.

This role sits at the intersection of data science, payments domain expertise, and production ML engineering. You'll partner closely with Fraud & Risk, Engineering, Compliance, and Product to turn data into models that run in production and not just notebooks that sit on a shelf.

Key Responsibilities

• Design, train, validate, and deploy ML models for use cases such as: real-time transaction fraud detection, card-not-present (CNP) risk scoring, authorization decline/approval optimization, chargeback/dispute prediction, merchant risk scoring, and AML/transaction-monitoring anomaly detection.
• Engineer features from transactional, behavioral, and device/network data while respecting strict latency budgets (often sub-100ms scoring at authorization time).
• Evaluate and select appropriate techniques - gradient boosting, deep learning, graph-based fraud detection, anomaly detection, time-series methods, based on the problem, not fashion.
• Work with Engineering to deploy models into real-time and batch pipelines, ensuring reliability, monitoring, and rollback safety in a payments-critical path.
• Build and maintain model monitoring for drift, data quality, and performance degradation, with alerting tied to operational and fraud-loss KPIs.
• Contribute to (or help establish) MLOps practices: versioning of models/features/data, reproducible training pipelines, CI/CD for models, and A/B or shadow-testing frameworks before full rollout.

AI/LLM-Adjacent Work

• Apply modern AI tooling - including LLMs such as Claude - to accelerate data science workflows: automated feature exploration, model documentation, anomaly narrative generation for fraud analysts, and code generation/review for data pipelines.
• Explore applied use cases for LLMs in risk and compliance narratives (e.g., summarizing suspicious activity patterns for SAR drafting support, with mandatory human review).

Cross-Functional Collaboration & Governance

• Partner with Fraud & Risk and Compliance teams to ensure model outputs are explainable and defensible to auditors, regulators, and card scheme risk teams (Visa, Mastercard).
• Present findings and model performance to non-technical stakeholders (Risk Committee, Product, Client Services) in clear, decision-useful terms.
• Ensure all data handling complies with PCI DSS, data residency requirements, and internal data governance policies - particularly around cardholder data (PANs, CVVs, authentication data).

Required Qualifications

• 10+ years of overall experience, including at least 3+ years as a Data Scientist or ML Engineer, ideally in payments, fintech, banking, or another environment with high-volume transactional data and real-time decisioning.
• Strong proficiency in Python (pandas, scikit-learn, XGBoost/LightGBM, PyTorch or TensorFlow) and SQL.
• Demonstrated experience building and deploying models into production (not just research/exploratory work), with attention to monitoring and retraining considerations.
• Solid understanding of classification, anomaly detection, and imbalanced-class problems (fraud is a classic rare-event problem).
• Experience with cloud data/ML infrastructure (AWS/GCP/Azure - e.g., SageMaker, Vertex AI, Databricks) and standard data engineering tools (Spark, Airflow, or similar).
• Understanding of the regulatory and security constraints of financial services data (PCI DSS, data minimization, access controls).
• Strong communication skills - able to translate model outputs and trade-offs (precision/recall, false positive cost, latency) into business decisions for risk and product stakeholders.

Preferred Qualifications

• Direct experience with payments-specific fraud typologies: CNP fraud, account takeover, first-party fraud, synthetic identity, BIN attacks, or card testing.
• Experience with graph-based or network analysis techniques for fraud rings/merchant collusion detection.
• Familiarity with card scheme rules and risk parameters (Visa Risk Manager, Mastercard Fraud attributes, or similar).
• Experience applying LLMs (e.g., Claude, GPT) to data science workflows - feature engineering assistance, automated EDA, report generation, or analyst-facing summarization tools.
• Exposure to real-time streaming architectures (Kafka, Flink) for low-latency scoring.
• MSc/PhD in a quantitative field (Statistics, Computer Science, Applied Math, Physics, Operations Research) or equivalent practical experience.