Vijay Bhatt

About Vijay Bhatt

I\'ve spent 19+ years building and scaling products across four fundamentally different domains — government IT, datacenter infrastructure, consumer search, and enterprise AI — proving that the best engineering leaders aren\'t domain-bound, they\'re problem-bound.Currently at Microsoft, I lead a team building enterprise AI agent experiences — multi-agent systems using LangChain, LangGraph, Claude, GPT-4.1, and MCP Server that operate within the highest enterprise security standards (tenant isolation, ALE, encryption at rest). Scaled to ~90K weekly active users with ~73% sustained satisfaction.Previously built Bing\'s health search vertical from zero to 0.5M DAU and $2M ARR across 60 countries, designing BERT/ColBERT/ANN-powered semantic retrieval pipelines.What I bring to every role: I build teams from scratch (three times, across two companies), I create evaluation-first quality cultures, and I turn ambiguous 0→1 ideas into globally scaled products.Key areas: Multi-agent AI systems, LLM orchestration, evaluation infrastructure, enterprise security, BERT/NLP, semantic retrieval, cross-org platform building.

Experience

1. Principal Engineering Manager

   Microsoft

   Jul 2025 — Present · Bengaluru, IN

   Leading 15+ member team building enterprise AI agents and the cross-org evaluation platform that gates every agent release.AI AGENT FOR ENTERPRISE COLLABORATION• Architected multi-agent system for enterprise collaboration — an AI teammate operating within collaborative workspaces, honoring membership privacy and security boundaries.• Designed agentic orchestration using LangChain & LangGraph — specialized agents coordinating via structured handoffs for project management, deadline tracking, automated reporting, and workflow automation.• Built multi-model AI stack (GPT-4.1, Claude, M365 Copilot). Developed Claude-native skills with MCP Server for tool-use orchestration, prompt routing, and model-capability-aware task delegation. Drove latency/quality tradeoffs via architecture reviews and A/B experimentation.• Shipped 10+ production workflows — Task Management, Status Reports, Project Management, Document Generation — replacing manual processes with autonomous agent pipelines.• Enterprise-grade security: tenant isolation, information-barrier enforcement, ALE, encryption at rest, Azure Managed Identity, zero-trust access. Own production reliability and live-site operations.EVALUATION PLATFORM — CROSS-ORG QUALITY INFRASTRUCTURE• Built end-to-end eval platform spanning offline batch eval, gating pipelines, online telemetry-driven eval, and satisfaction correlation — adopted across multiple agent teams and squads as the canonical release gate.• Defined scoring framework (groundedness, fidelity, aesthetics, consistency, completeness) with human-annotation loops calibrating automated metrics against real user satisfaction.• Caught multiple pre-production regressions in orchestrator toolchain and agent skills, preventing user-facing quality degradation. Championed eval-first quality culture across the agent portfolio.

Education

• Nirma University

  Master of Technology (M.Tech.), Computer Engineering

  2007 — 2009

• Sardar Patel University

  Bachelor Engineering (B.E.), Computer Engineering

  2001 — 2005

• St. Xavier's School

  HSC, Science Stream

  1990 — 2001

Skills

• Windows Phone/Xaml
• C/C++
• C# 4.0
• Web Services
• Pl/Sql
• C++
• Wpf
• Visual Basic
• Ssrs
• Asp
• C#
• Ssis
• Microsoft Office
• Vb6
• Msbi
• Vb.net
• Sharepoint
• Microsoft Sql Server
• Matlab
• Software Design
• Sql
• Asp.net