Poshmark — Scaling Social Commerce to 80M+ Users & $1.8B GMV

• Client: Poshmark · Redwood City, CA
• Engagement: Engineering Leadership & Scale Advisory
• Role: VP of Engineering (CTO of Appmetry)
• Stack: Ruby · Python · Java · Go · Node.js · MongoDB · Redis · Elasticsearch · Kafka · RabbitMQ · Kubernetes · AWS

Poshmark Scale at a Glance

• 📈 80M+ Registered Global Users
• 💳 $1.8B+ Gross Merchandise Value (GMV) processed
• ⚡ 99.99% High-Availability Platform Uptime
• 🔍 100M+ Listings searched with sub-second latency
• 🚀 10 → 800+ Org growth from early startup through IPO & acquisition
• 🔄 Millions of daily events (social follows, shares, likes) processed asynchronously

The Challenge

Poshmark is a social commerce marketplace where millions of users buy and sell fashion items daily. The engineering team faced a massive challenge: scaling the platform's core infrastructure from a ~10-person startup to a global marketplace supporting 80M+ users and processing $1.8B in GMV.

As the platform experienced hyper-growth, the existing monolithic architecture reached its limits:

• Developer Bottlenecks: Tightly coupled monolithic dependencies slowed feature releases.
• Database & Transaction Limits: High-frequency social actions (likes, shares, follows), search queries, and payment checkouts caused database lockups, especially during major holiday sales spikes.
• Search & Feed Latency: The search infrastructure struggled to deliver real-time, context-aware item discovery in sub-second speeds.
• Risk & Trust Overheads: Manual user moderation and fraud checks could not keep pace with transaction growth.

---

The Solution

Under the leadership of Ketan Jain (former VP of Engineering at Poshmark and current CTO of Appmetry), the engineering team executed a multi-year technical roadmap to decompose the monolith, modernize the payments platform, and integrate cutting-edge AI systems.

Key initiatives included:

1. Microservices Decomposition & Event-Driven Architecture

We refactored the core monolith into domain-driven microservices using Java, Go, Ruby, and Python to maximize development velocity. High-concurrency operations were offloaded to event-driven architectures powered by Kafka and RabbitMQ, with distributed caching handled by Redis and MongoDB. This transition enabled continuous delivery, increasing deployment frequency from bi-weekly releases to continuous, automated releases.

2. Global Payments Platform & Reconciliation

We designed and owned the end-to-end technology strategy for the core payments, payouts, reconciliation, chargeback, and shipping platforms. We integrated local payment methods and processors globally—including Adyen, Braintree, PayPal, Venmo, Sezzle, and Razorpay—to support localized checkout experiences, multi-currency processing, and secure payouts.

3. AI-Powered Personalization & Market Intelligence

We pioneered two major AI-driven platforms:

• AI Feed Personalization: Implemented intelligent ranking and recommendation systems using machine learning to increase buyer engagement, feed relevance, and marketplace conversion across millions of daily active users.
• AI Voice-of-Customer (VoC) Platform: Architected an AI-powered intelligence platform that utilized LLMs to consolidate signals from App Store reviews, Reddit communities, social media, and customer support tickets. This platform converted unstructured feedback into real-time, data-driven product insights.

4. Trust, Safety & Regulatory Compliance

We designed and scaled a multi-signal fraud, spam, and user moderation platform to protect marketplace integrity. Key security implementations included:

• Integrating SOCURE for automated KYC and identity verification.
• Implementing strict operational controls and secure data handling to achieve and maintain compliance for PCI DSS, GDPR, and CCPA.

5. International Expansion

We led the engineering efforts for successful launches in Canada, Australia, the UK, and India, managing complex localization, international compliance, and local vendor integrations.

---

The Impact

The modernization successfully prepared Poshmark's platform for its IPO and eventual acquisition by Naver:

• Scale Accommodated: Seamlessly supported 80M+ users and $1.8B GMV with 99.99% operational uptime.
• Deployment Speed: Releases shifted from a slow, risk-heavy bi-weekly process to continuous, automated canary deployments.
• Observed Latency: Item discovery and search indexing latency dropped to sub-second levels.

---

Why Global Startups & SMEs Trust Appmetry

Appmetry is not a typical agency. The exact architectural pedigree, high-concurrency database optimizations, global payments systems, and compliance rigor that scaled Poshmark are built into our team's DNA.

Whether you are a global startup looking to build a high-performance prototype with clean architecture, or an SME needing to modernize legacy monoliths, integrate AI workflows, or roll out localized billing systems, Appmetry delivers production-grade solutions you can scale with confidence.