Skip to content
harsh sundani
Portrait of Harsh Sundani

Data & AI leadership · Sunnyvale, CA

Harsh Sundani

Data & AI Executive. Director, Data & Analytics at LinkedIn.

I build data functions from nothing and turn them into the infrastructure a business runs on.

LinkedIn hi@hsundani.com Google Scholar Résumé (PDF)

LinkedIn · Eaton · Cypress Semiconductor · Cree · Vanderbilt MBA

01 / summary

Three times I’ve built one from nothing: Cypress, Eaton, and go-to-market at LinkedIn. A fourth is underway now, the finance data function for LinkedIn’s CFO.

At LinkedIn I lead a 35-person analytics and data science org. I hired and grew it, including the four managers who run it today. It serves 4,500+ people across Sales, Marketing, and Customer Success, and I set data and AI strategy with the CRO, the CFO, and the SVP of GTM Operations.

The systems are in production and used every day. A Text-to-SQL agent that fields 60%+ of 20,000+ questions a year. A customer-research agent. Churn and propensity models. An income-statement narrative generator for the finance close.

Before LinkedIn, I built Eaton’s sales analytics function from a team of one into a platform 1,200+ people used every day.

What I build gets used, not just built. Every function I’ve built outlives the person who built it.

02 / impact

$19M

at-risk revenue protected with production churn and propensity models

65,000+

GTM hours saved a year by the customer-research agent I built, letting reps cover about 20% more accounts

5 → 80%

executive QBR coverage, from an AI reporting engine

~75%

less data-architecture complexity, from one governed layer

03 / experience

LinkedIn

Director, Data & Analytics

Go-to-Market and Finance

since 2022

  • Built the data function from nothing into a 35-person organization across data engineering, data science, analytics, and AI, which I lead through four managers. It is now the default insights engine for 4,500+ people across GTM and Finance on a $4B SaaS book.
  • Shipped LinkedIn’s first production generative-AI agents for revenue and finance. One capability now ships inside Microsoft Sales Copilot.
  • Consolidated fragmented, domain-specific data into a single governed architecture. Complexity fell about 75%, which made self-serve, AI-powered analytics possible.
  • Promoted to Director in 2024. The CFO extended my charter to Strategic Finance in 2026.

Eaton

Director, Business Intelligence & Analytics

2016 to 2022

  • Served as Chief of Staff to the SVP of Sales Operations.
  • Founded the sales analytics function from nothing and grew it to a team of 12, supporting a $5B book of business.
  • Launched a self-serve analytics platform used daily by 1,200+ people, and moved the stack from on-prem SQL Server to Azure.
  • Deployed forecasting models that lifted average deal size 7% and rep productivity 20%.

Cypress Semiconductor

Senior Analytics Engineer

2013 to 2014

  • First data science hire for a $70M manufacturing operation. My models improved on-time delivery 45%+ and saved $1.5M a year.

04 / research and patents

3

granted U.S. patents

282

Google Scholar citations

5

h-index

250+

citations, lead survey paper

Lead author of Wireless Sensor Network Simulators: A Survey and Comparisons, cited more than 250 times. Peer-reviewed researcher and named inventor on three granted U.S. patents from earlier semiconductor and LED R&D at Cree.

Google Scholar →
granted patents US 9,287,475 → US 10,439,112 → US D709,464 →

05 / education

Vanderbilt University

MBA

Dean’s Scholar, full-tuition scholarship

University of Toledo

MS, Electrical Engineering

Research Scholar, full-tuition scholarship

University of Mumbai

BE, Electronics & Telecommunication Engineering

Graduated top of class

06 / writing

I write about data and AI strategy, agentic systems, and building teams that scale. I also coach data and analytics leaders on Leland.

There’s a small set of analysts I’ve hired… → Something I’ve been noticing in conversations… → During a recent conversation with a mentee… → There’s an old line in analytics, often attributed… → One thing AI-assisted analytics has surfaced… → When data quality degrades, the reflex is… → As a follow-up to my last post about getting… → One of the harder habits to break in analytics… →
See all posts on LinkedIn →

07 / contact

Let’s talk.

If you’re building a data or AI org, or trying to fix one, I’d like to hear about it. Email is the fastest way to reach me.

hi@hsundani.com → Connect on LinkedIn →