Abhinav Ankur

About Abhinav Ankur

I’ve spent a lot of time thinking about one specific problem: how do you take a brilliant AI prototype and turn it into a piece of industrial infrastructure that doesn\'t fall apart under pressure?Most people focus on the model itself, but I’m interested in the machinery underneath. As a Principal Architect at Verizon, my job is to build the engine—the actual \"AI Factory\"—that makes these systems safe, cheap, and fast enough for millions of people to use every day. We need systems that are observable, auditable, and built to last.What does that look like in practice?It means architecting a Unified SDK that handles the messy reality of multi-agent state management so other teams can just focus on building features. It means getting voice latency down under 500ms by colocating our inference with telephony data—because if the AI takes three seconds to think, the user has already hung up. It also means building things like Central Agent Registries, so if an autonomous system starts behaving in a way we didn\'t intend, we have the tools to pull it back immediately.Does this actually move the needle for the business?The numbers say yes. I recently architected an autonomous parsing system that was projected to save $17M annually by automating high-volume manual workflows—a massive operational burden we were able to solve through agentic design. Even as specific products evolve or sunset, the core takeaway remains the same: I look at ROI with the same rigor I apply to latency. If it isn\'t measurable, and if it doesn\'t solve a multi-million dollar bottleneck, it isn\'t working.I also believe that being a Principal means you\'re a force multiplier for the people around you. I currently lead several product pods and am mentoring over 20 engineers. There’s a unique professional satisfaction in helping a team bridge the gap between writing basic scripts and designing the kind of complex, multi-agent systems that can actually handle the scale of a Tier-1 carrier.

Experience

1. Principal Ai Ml Engineer

   Verizon

   Jun 2023 — Present · NJ, US

   Industrializing the AI Factory: Architected a Unified Agentic SDK built on LangGraph Enterprise and Google ADK to abstract multi-agent state management, moving the organization beyond fragmented prototypes to a production-grade kernel with \"streaming out-of-the-box\" capabilities.Standardized Connectivity & MCP: Launched native Model Context Protocol (MCP) integration to provide secure, standardized connectivity between LLMs and enterprise data silos (Jira, Slack, SQL), eliminating documentation silos and \"vibe coding\" patterns.Performance at Scale: Optimized the inferencing stack to achieve sub-500ms end-to-end voice latency by colocating model inference with telephony data halls and orchestrating parallel STT/LLM/TTS architectures.High-Throughput Infrastructure: Orchestrated the migration of high-traffic applications to the HA platform, managing 2M+ daily inferences and implementing \"dark testing\" strategies to validate performance under peak load without impacting production traffic.Sovereign Governance: Built a foundational \"Trust Layer\" featuring a Central Agent Registry and \"Reversible Autonomy\" toolkits to ensure high-stakes agentic actions remain observable, auditable, and retractable.Agentic Validation Ecosystem: Establishing a robust validation kernel for automated model evaluation, implementing Shadow Testing as a platform standard to validate agents against production-grade \"dark\" traffic.Technical Leadership: Lead multiple product pods in a player-coach model, managing the engineering \"Book of Work\" to align technical output with strategic business priorities and high-throughput digital infrastructure goals.

Education

• St. Michael's High School, Patna, Bihar

  Secondary School Certificate, Computer Science

  1999 — 2009

• Udacity

  Nanodegree, Machine Learning - Introduction with PyTorch

  2020 — 2020

• Lovely Professional University

  Bachelor of Technology - BTech, Computer Science

  2012 — 2016

• University of Illinois Urbana-Champaign

  Master's degree, Statistics

• St. Michael's High School, Patna, Bihar

  Higher Secondary Certificate, Computer Science

  2009 — 2011

Skills

• C# 4.0
• Ibm Streams
• Css3
• Microsoft Office
• Oracle Database
• Xaml
• Data Structures
• Css
• Powerpoint
• C
• Programming
• Drupal
• Sql
• Linux
• Microsoft Word
• Ajax
• Data Analysis
• Xml
• .net Framework
• C#
• .net
• Laravel
• Awk
• Machine Learning
• Web Development
• C++
• Spring Framework
• Html5
• Visual Basic
• Javase
• Javascript
• Software Development
• Python
• Php
• Bootstrap
• Cascading Style Sheets (Css)
• C++ Language
• Html
• Flash Professional
• Java