ratul_esrar
Booking projects for Q3 2026 · Los Angeles

Decision-ready analytics for healthtech and marketplace startups.

Your team is making calls without a shared picture of the data. I turn ambiguous questions into evidence-backed narratives your leadership can act on, and build the models underneath them. No full-time hire required.

Book an intro call See selected work
raw_sources · 8+ years across
TikTokCareRevEngineCorner HealthUChicago
stg_services

Three ways to work together.

Scoped engagements with clear deliverables, built for teams that need answers now and a foundation that outlasts the project.

01 · sprint · 2–4 weeks

Analytics Foundation Sprint

For lean teams without a dedicated data architect. I use AI-native tooling to trace data lineage fast and stand up a warehouse and dbt layer your team and your AI tools can actually trust.

  • A warehouse your team can actually query
  • Layered dbt models with tests + docs
  • A data model your AI tools trust as much as your analysts do
02 · project · 3–6 weeks

Product & Experimentation Setup

For product teams shipping on gut feel. I design your event tracking and A/B testing practice so every launch produces a readable answer.

  • Event tracking design (Amplitude / Pendo)
  • An A/B testing practice where every launch produces a readable answer
  • Funnel, retention & cohort analysis
03 · retainer · monthly

Fractional Analytics Partner

Your senior analyst, a few days a week. Ongoing KPI monitoring, root-cause investigations, and exec-ready readouts as your business changes.

  • KPI monitoring & metric deep-dives
  • Scenario & tradeoff modeling
  • Leadership-ready recommendations
fct_selected_work

Work that moved a number leadership cared about.

Specific figures shared in conversation, not on the open internet. Here's the shape of what I do.

Corner Health · analytics engineering, AI-native build · 2026

Compressing a quarter of data engineering into three weeks, solo.

Corner Health's 3-person engineering team had no backend engineer or data architect. Claims adjudication data lived in an external payer API, and the operational foundation connecting it to provider data was tracked by hand in Google Sheets.

I used Devin, the team's AI software engineer, to trace data flows through their codebase and map lineage from the claims API to their production provider data. From there, I built a layered dbt project (staging → facts/dims → metrics) to replace the manual spreadsheet with a real semantic layer. Notion AI turned working sessions with the co-founder into clear requirements and a statement of work; Basedash, an AI-native BI tool, powered the exec dashboard and operational drilldowns on top.

quarter → 3 weeks
typical timeline for this scope, using an AI-native build workflow
1 person
no dedicated data architect or backend engineer required
TikTok · Live revenue operations · 2024–25

Finding and keeping the users who drive livestream revenue.

Top livestream contributors were hitting payment friction, and high-value users were churning silently. I built TikTok Live's behavioral segmentation and risk scoring models, then worked with GTM and Finance to turn model outputs into proactive intervention protocols.

revenue ↑
double-digit lift from top livestream contributors
churn ↓
measurable retention gains among at-risk, high-value users
figures available on request
CareRev · healthcare staffing marketplace · 2021–24

Pricing a two-sided marketplace with models, not guesswork.

Shift pricing was set by intuition, leaving fill rates and margin on the table. I built scenario-based predictive pricing models, productionized them with the ML team on AWS, and built the dbt semantic layer that let product and ops answer their own questions without waiting on data.

fill rate ↑
more shifts covered at lower staffing cost
insight ↑
self-serve semantic layer cut time-to-answer for every team
figures available on request
int_the_lab

The lab: analytics applied to things I love.

Public demos of how I work: same modeling discipline, lower stakes. Real data, real findings.

Genre Crossover Map
interactive · hover the tracks · real Spotify data

Every dot is a real track, plotted by tempo and energy. Ringed dots are bridge tracks: songs that sit in a neighboring genre's pocket and can carry a set across genre lines. House and techno share almost the same space. Hip hop and drum-and-bass are mirror images: different ends of the tempo spectrum that meet in the same pocket, which is why the records that cross between them are the ones worth knowing.

demo repo · dbt + duckdb · healthtech

Claims Analytics Demo

A miniature revenue-cycle stack: synthetic claims through layered dbt models into payer denial-rate and cash-cycle metrics, the pattern from my Corner Health engagement, rebuilt in public.

view repo →
demo repo · python · music analytics

Genre Crossover Map

Finding the tracks that bridge worlds. Five genres mapped in tempo and energy using real Spotify data. House and techno share a pocket. Hip hop and drum-and-bass are mirror images: same tempo math, opposite directions. The bridge tracks between them are the ones a DJ builds a crossing around.

live demo → view repo →
peer-reviewed publication · ML for public policy

Predicting post-release interactions with mental health systems

Random forest models with temporal cross-validation to identify individuals at elevated risk after release. This research supported better reintegration outcomes.

read paper →
applied ML · privacy tooling

CamBurglar: detecting hidden Wi-Fi cameras

Classification and spatial modeling to flag unauthorized surveillance devices on wireless networks, shipped as a public-facing Flask app.

view repo →
dim_about

Analyst first. Engineer when it helps.

I've spent 8+ years turning ambiguous business questions into analyses leadership can act on, at a public company's scale (TikTok), through marketplace hypergrowth (CareRev), and inside scrappy healthtech teams (Corner Health).

My background is computational public policy at UChicago, which is why I gravitate toward work where the numbers matter beyond the dashboard: healthcare access, fair pricing, systems that treat people well.

Off the clock: open-format DJ sets, festival season, and cooking things that take too long on purpose.

  • baseLos Angeles, CA
  • experience8+ years, senior analytics
  • educationUChicago, M.S. Computational Analysis & Public Policy
  • stackSQL · Python · dbt · Snowflake · BigQuery · Looker · Amplitude
  • domainsHealthtech · Marketplaces · Product & Revenue
  • engagementSprints, projects & fractional retainers
mtc_contact

Have a data question your team can't answer this week?

Tell me what decision you're trying to make. If I'm not the right fit, I'll say so and point you to someone who is.

ratulesrar@gmail.com Schedule 30 minutes
github linkedin