The Challenge
Most credit systems in India were built for salaried workers with clean paperwork. That leaves out the majority: street vendors, gig workers, small shop owners. The people who most need credit are the ones the system can't evaluate.
My mission at HyperVerge was to lead the Modules charter and make credit underwriting work for everyone, not just the people with payslips and credit histories.
What I Built
Three years, three phases. Each one started with a blank slate and ended with something that shipped.
Phase 1: Platform Foundation (2023)
Joined a new 3-person team inside the company. No roadmap, no playbook. Built the building blocks.
-
API Orchestration Layer
Integrated 85+ third-party verification APIs into a single platform. Reduced deployment time by 50%.
-
HyperVerge One Platform
Core infrastructure for the product suite. Built APIs, workflows, sales decks, and client demos. Wore every hat the role demanded.
Phase 2: Market Expansion (2023–2025)
Took those building blocks to market. Made products sellable, partner-ready, and scalable.
-
Signature Detection
Identified a blue-ocean opportunity most teams avoided due to complexity. Built a proprietary AI model that outperformed Amazon AWS in accuracy. Opened a new revenue stream.
-
Bank Account Verification
Reduced turnaround time from 5 minutes to 10 seconds using Reverse Penny Drop. Made real-time account verification viable at scale.
-
API Marketplace
Made products partner-ready. Achieved 3x funnel conversion lift and reduced product integration time from 14 days to 5.
Phase 3: AI Underwriting Suite (2025–2026)
Promoted to PM. Led a 9-person cross-functional team to build the credit underwriting products.
-
Financial Document OCR
AI-powered extraction from bank statements, ITRs, and GST returns with 95%+ accuracy. Reduced manual review from hours to seconds.
-
Account Aggregator Integration
India's first seamless AA-based data pipeline for real-time financial profiling, enabling lenders to pull verified data with user consent.
-
Low-Code Onboarding Platform
A drag-and-drop workflow builder that let lenders configure underwriting rules without engineering support. 4x increase in average revenue per transaction.
By reducing Turnaround Time from several hours to under 3 minutes, we enabled rural shop owners to get microloans almost instantly, a process that previously took days of paperwork and branch visits.
"Aayush stands out for his rare combination of curiosity, clarity, and technical empathy... He consistently steps beyond the formal boundaries of his role and engages deeply with the technical side of AI products."
— Narasimha Y, Lead ML Engineer at HyperVergeKey Achievements
-
0-to-1 Foundation
Built the platform foundation as one of 3 team members, shipping core modules that still form the product backbone two years later.
-
Revenue Impact
Scaled 3 major modules to Product-Market Fit, generating $1.2M ARR and driving 0→1 revenue in India.
-
Operational Efficiency
Reduced product integration time by 65% (14 → 5 days), achieved 70% Straight-through Processing, and cut account verification TAT from 5 minutes to 10 seconds.
-
Infrastructure Overhaul
Redesigned backend infrastructure to support 85+ verification APIs, reducing deployment time by 50% and enabling rapid client onboarding.
-
Team Leadership
Grew from a 3-person team to leading a 9-person cross-functional team across engineering, ML, design, and QA.