Mohit Narang

About Mohit Narang

The most dangerous AI system isn\'t one that breaks. It\'s one that runs unsupervised in a regulated industry — and no one can explain what it did, or why.I\'ve spent 20+ years watching smart people make expensive mistakes. Not because we lacked intelligence — but because the systems around us didn\'t capture it, govern it, or carry it forward. That\'s what drew me to AI, not just the automation or efficiency gains. The possibility of building systems that preserve human judgment — that know when to pause, and wait for a person to say: yes, proceed.I recently built two open-source systems that put this into practice:Gene — governed Multi-Agent System with schema-first architecture, a mandatory Human-in-the-Loop gate, and a full JSON audit trail a compliance officer can actually sign off on.LoanIQ — A2A credit intelligence system where two agents negotiate incomplete financial data in real time, governed by hard policy as code. Intake to decision. Under a second. Complete audit trail.What I bring to Transformation leadership:→ 20+ Yrs FinTech Delivery — $15M+ portfolios, agile teams, client relationships end-to-end, deep fluency in fintech workflows, regulated environment delivery, HIPAA/PCI DSS-aware design→ Enterprise AI Strategy — Governed, executable roadmaps for regulated environments → Architecture Credibility — I build agentic systems myself. LangGraph, stateful multi-agent pipelines, inter-agent negotiation protocols, and HITL governance design.→ Governance as Advantage/ Governed AI — policy as code (not prompt), Pydantic schema enforcement, explainable AI, compliance-ready audit lineage → AI Observability — Custom AgentOps layers tracking E2E trace duration, A2A handoff latency, guardrails violation rate, factual accuracy, first-pass approval rate, and 6 more→ LLM Integration — Prompt engineering, strict JSON discipline, RAG with retrieval precision tracking, hallucination detection in production. → Cloud & MLOps — Azure, CI/CD, GitLab, Harness, DevOps for AI workloadsTechnology doesn\'t transform organizations. People with the right tools do.MIT — Applied Agentic AI for Organizations TransformationGene · LoanIQ — Open Source on GitLab

Experience

1. Sr Director

   Fiserv

   May 2025 — Present · NJ, US

Education

• Mahatma Jyotiba Phule Rohilkhand University

  Bachelor of Technology - BTech, Computer Science & Information Technology

Skills

• Customer Relationship Management (Crm)
• Business Analysis
• Software Development
• Ms Crm 2011
• Agile Methodologies
• Project Planning
• Team Leadership
• Microsoft Sql Server
• Biztalk
• Vendor Management
• Software Project Management
• Crm
• Microsoft Crm
• Eai
• Stakeholder Engagement
• Program Management
• Sdlc
• Solution Architecture
• Manage Client Relationships
• Offshore Software Development
• Requirements Analysis
• Ms Project
• Testing
• .net Framework
• .net
• Resource Management
• Web Services
• Project Management
• Product Development
• Product Requirements