Vinit Somani

Built 0 1: From scratch to $10M+ DEX volume, 10K+ users, 2K+ active traders in 6 months. Shipped AI-ranked Smart Money Feed: Using wallet clustering and LLM 2.5 DAU lift in 60 days, 2.3 increase in active traders. Referral program, KOL partnerships, and Telegram reactivation: +9% signups, +16% acquisition traffic, +22% returning sessions. Integrated non-custodial wallet infra: Embedded signing account abstraction reduced onboarding friction from 4 steps to 1-tap. Architected trading execution UX: Built copy-trading & limit-order infrastructure with fee-preset modules. Social analytics layer: WebSocket streaming 24hr tweet volume, sentiment scoring, hype gauges, and notification systems. Analytics stack: Event taxonomy, funnel dashboards, cohort tracking achieved 9+ minute avg session duration, 10%+ retention.

Defined and shipped a trustless AI verification layer to address hallucinations in LLM-driven workflows. Designed an end-to-end model claim verification consumer pipeline: Enabling claim-level provenance for AI outputs. Architected core AI platform components: Claim Engine converting LLM outputs into structured, machine-verifiable claims. Verifier Network with configurable trust assumptions. Multi-agent validation with confidence scoring. Evaluation layer for hallucination. Improved AI output reliability by ~15%: Enabling safer downstream consumption across enterprise use cases. Built a 0 1 B2B HRTech SaaS product on above model: Including multi-vendor ingestion, deduplication logic, SLA tracking, and RBAC.

Built multi-chain mobile and web wallet: Seed/login UX, portfolio views, transfers, chain/network considerations. Led engineering efforts: On wallet-core APIs, signing flows, and IBC/bridge compatibility, ensuring secure, extensible infra. Defined and launched Leap Elements SDK: A developer toolkit enabling the embedding of trading directly into front-ends. Feedback loops: 50+ user interviews across developers and communities, surfacing SDK and wallet UX friction roadmap iterations. Designed compliant onboarding and payment workflows: KYC/KYB, funding, settlement. Built event-driven ledger and reconciliation logic: Supporting cross-border payment flows. Dashboards defining NSMs: Activation, retention, transaction success) to prioritize features based on engagement impact.

Built platform primitives: Object models, workflow engines, rule-based automation systems, and RBAC-driven access controls for enterprise finance operations; contributed to the definition of enterprise reconciliation workflows for intercompany AR/AP use case. Designed bank reconciliation automation pipelines: Rule-based engines and custom bank-file parsers 80% reconciliation automation across R2R workflows; translated manual Excel-heavy finance processes into structured workflows enabling ML model. Transformed manual finance operations to: Structured, machine-readable workflows, enabling downstream AI/ML integration.

Improved AI-based cash-forecasting model accuracy by 8%: Tuned NLP extraction, dispute prediction algorithms, and customer-specific calibration; developed AI-assisted reconciliation systems with evaluation logic and edge-case handling. ERP and banking integrations across multiple enterprise deployments, reducing onboarding friction and accelerating time-to-value.