Kartik S H - Portfolio

Contributor Platform

Evaratus
In production ( 3 Weeks )

Team

Naman Bhalla ( PM )
Shreyansh S ( Manager )

Note : To comply with my NDA, I have omitted and obfuscated confidential information in this case study. All information presented is my own and does not necessarily reflect the views of Scaler.

Note: To protect confidential information, all product data and identifiers have been modified. This case study reflects my work during my time at Scaler.

Background

Evaratus, an initiative by Scaler, connects domain experts with AI training projects for frontier labs and enterprise teams. Contributors earn payouts by completing tasks like data creation and response evaluation.

Context

Payouts are the primary trust anchor on Evaratus, yet under 15% of contributors could independently track what they were owed across bi-weekly cycles. A large chunk of operational bandwidth went into answering routine payment questions and manually verifying earnings for confused users.

The low payment clarity and manual support burden highlighted the need to overhaul payouts and the linked banking setup.

Context

The low payment clarity and manual support burden highlighted the need to overhaul payouts and the linked banking setup.

Contributors

Subject matter experts across software engineering, healthcare, research, writing and more are matched to projects based on their specific background.

Insights

Early tool was built for low-volume testing, not parallel enterprise workflows. Proactively establishing scalable UI and data patterns at this stage was critical to support expanding project volumes while preventing compounding design and engineering debt down the line.

Here are few core findings from our internal contributor and operations research —

End-of-Cycle Task Flooding: At the close of each settlement, the payout view flooded with hundreds of unbundled task entries. Expanding users across multiple simultaneous projects made it impossible to manually cross-check completed work against deposited amounts.
Invisible Deductions & Rejections: Because in-app totals lacked line-item context for deductions or rejected tasks, users were forced to download raw settlement files to audit their earnings—eventually contacting support when numbers still didn't add up.
Inflexible Settings Structure: The account setup relied on fragmented, one-off UI patterns that couldn't scale. It lacked the structural foundation needed to support clear navigation, upcoming security verification, or future integrations.

Insights

Here are few core insights from our internal contributor and operations research —

⍟ End-of-Cycle Task Flooding: At the close of each settlement, the payout view flooded with hundreds of unbundled task entries. Expanding users across multiple simultaneous projects made it impossible to manually cross-check completed work against deposited amounts.
Invisible Deductions & Rejections: Because in-app totals lacked line-item context for TDS deductions or rejected tasks, users were forced to download raw settlement files to audit their earnings—eventually contacting support when numbers still didn't add up.
Inflexible Settings Structure: The account setup relied on fragmented, one-off UI patterns that couldn't scale. It lacked the structured foundation required to support upcoming security checks or any future payout integrations.

Understanding Current Scenario

Once a contributor submits work, reviewers assess the output against quality criteria. Successful submissions accumulate over a bi-weekly window, and the finalized total is disbursed directly to the contributor's registered bank account at the close of the settlement cycle.

Payouts(1) and Settings(2) before the redesign.

Goals of design

A frictionless payout experience is not only beneficial to contributors, making their earnings transparent and reliable, it's also critical for both operations and Evaratus.

For Contributors (Domain Experts) :

✅ Reconcile bi-weekly pay without manual spreadsheet audits.

For Operations & Support Teams :

✅ Eliminate manual end-of-cycle calculations and audit sheets

✅ Clean, standardized ledger states simplify cross-referencing banking UTRs against platform records.

For Evaratus & Scaler AI Labs :

✅ Retain top domain experts to deliver enterprise-grade AI datasets

✅ Build a scalable, SOC 2/compliance-ready financial foundation

⚠️ Risk of contributor churn and operational bottlenecks if payments break trust

Moving from low-volume testing to enterprise-scale required mapping every state change between our ledger and external banking rails. This exposed the gaps where asynchronous API delays and unbundled multi-project earnings were creating blind spots for both side of users.

Usage Scenarios

Usage Scenarios