AI strategy · platforms · research

Shriniwas
Suram

I lead AI strategy and agent platforms at U.S. News.

Academic Insights AI is the company’s first revenue-generating GenAI product, now on a platform of 20 million people. Independent research covers Mixture-of-Experts serving at the edge, information governance, and technology policy.

See the workRead the papersOpen to conversations
Abstract field of light representing large-scale AI interactions

0M+

People on the platform

$0M

ARR, Academic Insights AI

0K+

Paid subscribers

0%

Lower LLM inference cost

Selected work

Production GenAI at U.S. News, plus open-source agent infrastructure.

Laptop on a studio table showing an AI insights product interface

U.S. News & World Report

Academic Insights AI

The company’s first revenue-generating GenAI product. A multi-agent system across 40+ APIs, with MCP, guardrails, memory, observability, and evals — now at 50K+ paid subscribers and ~$4M ARR.

$4M ARR50K+ subscribers20M+ users
Minimal architectural doorway suggesting a secure gateway

Open source

Agent Gateway

A contract layer for agent discovery, custom protocols, and audited actions — so LangGraph, A2A, and MCP agents can call each other without giving up control.

MCP + A2AFull audit trail
On-premise GPU workstation and server rack in a pale studio

Platform

Hybrid Inference

A routing layer that keeps routine work on local models and bursts complex queries to the cloud. Fine-tuned Gemma and Qwen on-prem, with Vertex AI and OpenAI as overflow.

40% lower cost1TB+ / day
A single glass module floating above a studio table

Open source

Outcome as a Service

One-API templates for SLA-backed agentic workflows. Graph, evals, tracing, and safety bundled so teams ship a monitored outcome in hours, not weeks.

Hours to deploySLA-backed
Glass orbs connected by luminous threads in a studio

Greedy AI

Autonomous Agents

LangGraph agents with persistent memory, circuit breakers, and human-in-the-loop gates. Company-wide standard. Zero critical production incidents from automated tasks.

Sub-second−50% incidents

Research

Mixture-of-Experts serving at the edge.

Independent work on Mixture-of-Experts serving at the edge, plus PhD research on information governance and technology policy.

Google ScholarORCID 0009-0009-0452-9407
Close-up of a GPU with light traces suggesting expert routing

arXiv preprint · 2026 · arXiv:2608.18261

Cacheable by Design? Training Mixture-of-Experts Routers for Locality Against the Edge Memory-Bandwidth Wall

Serving a 235B MoE on a single 8 GB GPU is bound by memory bandwidth, not compute. Training routers for cache locality can cut misses; stacked with training-free cache-aware rerouting, miss reduction reaches ~80% at ≤3.4% perplexity.

arXivPDFCodeDOI
Translucent documents and a globe of light on a studio table

SSRN · 2026

Information Governance and Compliance Strategies for Data Management in the IT Automation Landscape

How AI and automation intensify the volume, velocity, and variety of enterprise data — and the governance and compliance strategies needed to keep automated systems accountable.

SSRNDOI
Crystal globe and policy still life on a pale studio table

SSRN · 2025

Emerging Technologies and Policy: Blockchain, Big Data, and the Role of Government in the Global Economy

The convergence of blockchain, big data analytics, and government policy as a defining force in the architecture of the contemporary global economy.

SSRNDOI

Experience

  1. Sep 2024 — Present

    Washington, D.C.

    Senior AI Software Developer & Researcher

    U.S. News & World Report

    • Launched Academic Insights AI — the company’s first revenue-generating GenAI product. 50K+ paid subscribers, ~$4M ARR.
    • Defined AI strategy with the C-suite and built a unified platform for agentic solutions across product lines.
    • Architected multi-agent systems across 40+ APIs, with MCP, guardrails, memory, observability, and evals, serving 20M+ users.
    • Built query-driven model routing with context caching, cutting monthly LLM inference cost by more than 50%.
    • Co-founded the Tech Center of Excellence and mentored engineers to shorten delivery cycles.

  2. Feb 2024 — Sep 2024

    San Francisco

    AI Engineer

    Greedy AI

    • Architected LangGraph agents with persistent memory to resolve multi-step queries at sub-second latency.
    • Moved the full-stack UI to TypeScript, React, and GraphQL, raising weekly developer velocity by 25%.
    • Engineered MLOps pipelines with anomaly detection and lineage tracking.
    • Co-authored a research proposal for BAA-funded Army Research Laboratory AI work.

  3. Jul 2022 — Nov 2023

    New York

    Data / ML Engineer

    Forbes

    • Built deep-learning propensity models on 1PB+ of data with TensorFlow to predict subscription conversion.
    • Shipped high-throughput inference microservices against millisecond SLAs.
    • Deployed a multi-turn LLM chatbot with safety guardrails, lifting engagement and retention.

  4. Jul 2021 — Dec 2021

    New York

    Cloud DevOps Intern

    Forbes

    • Automated GCP/AWS deployments with disaster-recovery and fault-mitigation protocols for infrastructure as code.
    • Stood up telemetry and alerting for high availability.

About

I build and lead high-impact AI products, including launching Academic Insights AI at U.S. News & World Report—its first GenAI revenue product, now serving 20M+ users. My work spans agentic frameworks at Greedy AI, ML platforms at Forbes, and independent research on Mixture-of-Experts edge serving and enterprise AI governance. Currently, I'm a PhD student focused on the operational risks and adoption of AI in organizations.

Education

  • PhD in Information Technology

    University of the Cumberlands

    Cybersecurity and artificial intelligence. Operational impact, adoption, and risk of enterprise AI.

  • M.S. in Information Technology and Management

    The University of Texas at Dallas

  • B.E. in Mechanical Engineering

    Pune University

Capabilities

Models, agents, and infrastructure.

AI systems

LLMs & LVMsLangGraphMCP / A2ARAGNeMo GuardrailsEvalsRLHFQuantization

Infrastructure

KubernetesvLLM / TensorRTVertex AISageMakerTerraformAirflowKafkaSpark

Languages

PythonTypeScriptSQLGraphQLReact

Cloud

GCPAWSAzureBigQueryPostgreSQLRedis

Certifications

  • TensorFlow Developer

    Google

  • Machine Learning Engineer

    Google

  • Data Engineer

    Google

  • Solutions Architect

    AWS

  • Machine Learning

    AWS

Contact

Get in touch.

Open to leading AI work, research collaboration, or a short note.

shriniwassuram@gmail.comLinkedIn GitHub