Senior Data Scientist

Company: ITTStar Global Services Pvt Ltd

Website: Visit Website

LinkedIn: Visit LinkedIn

Business Type: Small/Medium Business

Company Type: Product & Service

Business Model: B2B

Funding Stage: Bootstrapped

Industry: Information Technology

Salary Range: ₹ 15-25 Lacs PA

Job Description

We are seeking a Senior Data Scientist specializing in Agentic AI to design, build, and deploy production-grade autonomous multi-agent systems across regulated Fintech and Healthcare environments. The role involves developing intelligent agent workflows for fraud detection, credit underwriting, clinical decision support, and medical summarization, while ensuring compliance, explanability, and operational reliability.

Experience: 7–9 years | Domains: Fintech + Healthcare (both mandatory)

• Design and deploy production-grade autonomous multi-agent systems using LangGraph, AutoGen, and CrewAI for fraud detection, loan underwriting, patient risk stratification, and clinical summarization workflows.
• Build tool-using LLM agents that autonomously invoke credit bureau APIs, FHIR patient records, sanctions databases, and drug interaction databases via MCP (Model Context Protocol) and structured function calling.
• Engineer production RAG pipelines grounded on RBI/SEBI regulations, PCI-DSS policies, HIPAA guidelines, and clinical protocols - enabling agents to cite evidence and explain decisions to auditors and clinicians.
• Enforce structured, schema-validated agent outputs using Instructor or Outlines for reliable integration into downstream financial transaction and clinical EHR systems.
• Build real-time feature pipelines on Kafka and Flink feeding agent decision loops with sub-100ms latency for fraud alerting and patient deterioration detection.
• Develop and maintain classical ML models -XGBoost, LightGBM, Cox PH survival analysis — with full SHAP and LIME explanability reports for RBI/SEBI risk committees and clinical governance boards.
• Implement persistent agent memory using Mem0, Zep, or vector stores (Pinecone, Weaviate, pgvector) and design self-critique reflection loops for output quality and compliance validation.
• Fine-tune foundation models using QLoRA and PEFT on proprietary fintech transaction data and de-identified clinical notes for entity extraction, classification, and report generation.
• Instrument full agent traces using LangSmith or Langfuse; own CI/CD pipelines via GitHub Actions and MLflow with automated regression testing before every production deployment.
• Implement responsible AI guardrails using NeMo Guardrails and PHI de-identification pipelines via Microsoft Presidio ensuring HIPAA, DISHA, and PCI-DSS compliance across all data flows.
• Monitor model and agent drift using Evidently AI or NannyML; design agents that autonomously trigger retraining workflows or human escalation when confidence thresholds are breached.
• Partner directly with Chief Risk Officers, compliance teams, ICU physicians, and hospital administrators to translate domain policies and clinical hypotheses into agent workflows.
• Hands-on expertise required across LangChain, LangGraph, AutoGen, PyTorch, HuggingFace, Kafka, Spark, dbt, Kubernetes, Ray Serve, AWS SageMaker, and GCP Vertex AI.
• Deep domain fluency mandatory in fintech risk frameworks (fraud, AML, credit) and healthcare data standards (HL7 FHIR, OMOP CDM, Clinical NLP, FDA SaMD).
• Must have at least one autonomous agentic system shipped to production in a regulated environment; ability to mentor junior data scientists and define team-wide agentic AI standards expected.