Rushi Sheth — Systems, People & AI

Quantified impact

GTM & Commercial

$110M revenue growth

Sub-national ML + rules targeting unlocked $110M in incremental revenue.

$185M forecast defended

Fixed journey rules, preventing a $185M annual revenue overestimation.

1.5K → 11K Rx/week

Owned brand analytics and executive MBR/QBR readouts as weekly new customers grew from 1.5K to 11K, reaching $1.7B in annual revenue.

3% incremental sales

CRM analytics reallocated field capacity, creating 3% incremental sales.

60M patient TAM mapped

Productized a 60M-patient market map with funnel, growth, and segmentation views informing commercial strategy.

Omnichannel analytics

Touchpoint attribution model identifying highest-reach and highest-conversion channels across the media mix.

Coverage cadence insight

Cohort analysis revealed call density predicts provider activation — 44% of activated providers reached meaningful depth only with 9+ months of sustained monthly engagement.

Data Products

4-tier affiliation model

Built four-tier affiliation model powering enterprise-wide field targeting.

2 days → 2 hrs

Customer 360 refresh via PySpark/SQL lakehouse — batch bottleneck to near-real-time.

400K+ entities unified

Network Intelligence Tool mapping HCP affiliations into a single commercial decision layer.

Customer 360 dashboard

Tableau platform consolidating HCP profiles, patient panels, brand usage, affiliation networks, and insurance coverage.

AI / ML & Optimization

Enterprise Gen AI Pilot

Brand-level LLM deployment — translating commercial requirements into technical scope, validating semantic schemas, driving UAT.

Databricks AI workflow

Embedded AI-assisted debugging and prompt frameworks across a 7-analyst team — accelerating onboarding and reducing manual reporting time.

Customer at risk Classifier

ML classifier surfaced switch-risk customers using longitudinal customer behavior.

$12M media evaluated

MMM quantified omnichannel ROI, informing $12M quarterly media investments.

7% supply chain savings

CPLEX-based optimization reducing transportation costs by 7% across 500 customers.

AI + Workflows

Full writeups →

QueryMind

RAG + Text-to-SQL. Natural language over enterprise databases. dbt semantic layer, LangChain, vector store, 30-question eval suite.

↗ GitHub ↗ PRD

Agentic Booking

WhatsApp-native scheduling agent for service businesses. Claude tool-use loop: check plan → availability → preferences → book → confirm. No back-and-forth.

↗ Full writeup

Agentic HCP Targeting

Concept: multi-agent LLM workflow ingesting unstructured market studies to dynamically recalibrate provider target tiering — replacing manual, infrequent recalibration.

↗ Concept overview

Careway

Clinical pathway coordination — Patient, PCP, Specialist on shared live data. Designed through 6 iterations. Agentic layer: coordinator gets pre-call brief across pathway, overdue items, and outreach queue.

↗ App design + flow

Pharma Analytics · Transferable Expertise

Analytics built for complexity.

In pharma, I operate across two distinct customer types simultaneously — and the analytical infrastructure I build is directly transferable to any data-rich commercial environment.

The same framework, different domains The dual-customer model that maps HCP-Patient in pharma maps Host-Guest at Airbnb, Driver-Rider at Uber, Seller-Buyer at Amazon, and Coach-Member at WHOOP. In digital health, members replace patients, engagement data replaces claims, and the longitudinal journey from signal to outcome is the same analytical challenge at a different layer of the health stack.

Dual-Customer Commercial Analytics Framework

Customer 01

Healthcare Provider (HCP)

Affiliation Mapping400K+ entity network — where HCPs work, who they influence, referral patterns, practice affiliations.

Prescribing BehaviorRx patterns, market share by molecule, competitive dynamics, over/under-served territory detection.

Network IntelligenceInfluence mapping across HCP clusters to identify key opinion leaders and cascade targeting.

Integrated signal

Customer 02

Patient

Claims-Based Segmentation60M patient universe — 10 ML clusters defining the addressable market map and priority populations.

Longitudinal Journey MappingTracking diagnosis → treatment → switching → persistence over time using multi-million claim records.

Population TargetingIdentifying which patient segments to target, through which HCP channel, and at what point in the journey.

From data to commercial action

Raw Claims

Multi-million Rx & medical records ingested

›

Segment

ML clusters define addressable patient populations

›

Map Journeys

Longitudinal paths reveal when & where to intervene

›

Link HCPs

Affiliation network identifies the right prescriber

›

GTM Action

Field targeting, call plans, resource allocation

Longitudinal journey mapping is not easy — it requires linking millions of fragmented records across time, normalizing patient identities, and making sense of non-linear treatment paths. In pharma this means knowing which patient, through which physician, at which point in their disease journey to target.

The infrastructure I build — claims pipelines, affiliation databases, ML segmentation models — is the same architecture that drives customer intelligence in retail, fintech, and B2B commercial strategy.

Transferable to any industry

The underlying skills are universal

Longitudinal patient journeys → Member/rider/guest retention curves, churn prediction

HCP affiliation network → Host supply network, seller ecosystem, driver supply mapping

Claims segmentation (60M) → User cohort modeling at consumer scale, TAM sizing

Territory design → Market expansion, geo-based supply planning, coverage optimization

Omnichannel attribution → Performance marketing analytics, growth channel optimization

AI + Workflows

Building the next layer.

✦ Full project writeups →

Shipped product and in-flight concepts at the intersection of AI, GTM, and commercial operations.

Shipped · AI Product

QueryMind

RAG-powered, natural-language analytics over enterprise data. Business users query databases without writing a single line of SQL. Built end-to-end: schema embedding, semantic retrieval, query generation.

GitHub Repository Product Requirements Doc (PRD) System Architecture

Loom Demo — coming soon

In Dev · v0.2

rushi.ai — Portfolio Copilot

RAG-powered agent to chat with my full resume, projects, and context in real time — right in this browser.

✓

Knowledge base & embedding

→

RAG pipeline & agent logic

UI/UX integration

Memory & context agent

In Dev · Agentic

Pilates Scheduling Agent

WhatsApp-native booking agent. Customer says "book me something this week" — agent checks plan validity, availability, past preferences, picks the best slot, books it, confirms. No back-and-forth.

Full writeup + system diagram

Design · In Progress

Careway

Clinical pathway coordination — three roles (Patient, PCP, Specialist) on shared live data. Designed through 6 iterations. Agent layer: coordinator asks "prep me for my 2pm call" → agent synthesises pathway + overdue items + outreach brief.

App flow diagram + design notes

How I operate

The Value Loop

I operate across every stage — bridging the gap between technical infrastructure and executive commercial decisions.

01 · Signals

Sense

Customer behavior, operational data, market changes, stakeholder inputs.

02 · Structure

Model

Data models, pipelines, semantic layers, unified entity maps.

03 · Intelligence

Analyze

ML, segmentation, forecasting, optimization, AI workflows.

04 · Decisions

Prioritize

Targeting, resource allocation, GTM actions, executive alignment.

05 · Execution

Ship

Field action, ops changes, measured outcomes, loop back.

Toolkit

Skills & Infrastructure

Data Engineering & AI

dbtDatabricksSnowflakePythonPySparkSQLLangChainAgentic AIRAGText-to-SQL

Applied Analytics

Behavioral SegmentationLongitudinal AnalysisMarket Mix ModelingPredictive MLForecastingNetwork OptimizationOperations ResearchCPLEX / AMPL

Strategy & GTM

Program Management0-to-1 ScopingExecutive StoryboardingStakeholder ManagementGTM ExecutionTerritory DesignMarket Defense

Background

Where it started.

Education

Lehigh University

M.S. / B.S. Mechanical & Industrial Engineering. Led a nationally ranked Robotics team — systems thinking and constraint optimization before the first SQL query.

Operating Style

Cross-functional at heart

Designing cadence — dashboards, reviews, escalation paths — that keeps product, ops, marketing, sales, and leadership aligned and moving.

Formal Training

Beyond Engineering

Coursework in Managerial Economics, Econometrics, Financial Management, and Marketing — building intuition for causal reasoning, market structure, and the business model behind the data.

Off the clock

Constant learner

Cooking, photography, gardening. Fueled by podcasts spanning business, global markets, mythology, and history.

CookingPhotographyGardeningPodcasts

Let's build something that moves the needle.

If your business runs on understanding how people behave, grow, and churn — let's talk.

✉ Get in touch ↗ Connect on LinkedIn

Complex Data.People at Scale.Commercial Impact.