Agentic AI for Identity Verification
Autonomous multi-agent workflow orchestrating data validation, decisioning, and continuous learning. Hybrid rule-based + ML + agent reasoning for accuracy and explainability.
Staff Data Scientist building agentic AI systems for identity verification, fraud detection, and risk — at production scale, in regulated environments.
I lead data science work at Socure, where I design and ship machine-learning systems that verify identities and stop fraud for top US banks and fintechs.
My recent work has focused on agentic AI — orchestrating multi-agent workflows for autonomous decisioning, continuous-learning pipelines, and explainable risk models that meet FCRA and GDPR requirements.
Before Socure I spent five years at Conduent shipping NLP, computer-vision, and document-intelligence products across legal, healthcare, and ecommerce.
Autonomous multi-agent workflow orchestrating data validation, decisioning, and continuous learning. Hybrid rule-based + ML + agent reasoning for accuracy and explainability.
GenAI system translating SHAP-based feature importance into FCRA-compliant adverse-action notices. Multi-language, regulator-grade audit trails.
Fuzzy-matching agent with adaptive thresholds, NLP-driven name-variation expansion, and phonetic algorithms for high-recall sanctions screening.
Automated test-case generation and execution for ML products. Replaces manual QA pipelines with agent-driven evaluation, faithfulness scoring, and regression detection.
Multi-agent intake, routing, resolution, and QA pipeline using sentence-transformer embeddings for case similarity and priority-queue optimization.
ML/AI pipeline parsing hundreds of pages of contracts to surface monetary risk and savings opportunities. Action-oriented insights for legal and finance teams.
Vybhav consistently demonstrated the rare ability to translate complex ML research into production systems that moved the business — his work on identity verification directly impacted top-tier bank clients and the broader fraud roadmap.Engineering Leader Socure · Identity & Fraud
He doesn't just build models — he frames the problem, aligns the org, and drives outcomes. The agentic AI work he led was genuinely ahead of the curve, and he made it accessible to product and legal teams alike.Product Partner Cross-functional · Risk Products
One of the most thoughtful data scientists I've worked with. Strong technical depth across NLP, ML, and GenAI, paired with the patience to mentor and the clarity to present to executives.Senior Data Scientist Former Conduent colleague
Vybhav owned the analytics strategy end-to-end — from the dashboard that lifted KPIs by 35% to the pre-sales custom models that drove millions in pipeline. Rare combination of rigor and commercial sense.Senior Manager Socure · Customer Strategy
— More on LinkedIn recommendations ↗