As a Full Stack Engineering Intern at DMFlo, I’m working across the stack to build a production SaaS platform that automates Instagram DMs and comment-to-DM workflows for creators and businesses.
On the infrastructure side, I architected a webhook-driven ingestion pipeline that decouples inbound Instagram events from downstream processing, sustaining ~25 events/sec per instance using Redis and BullMQ. To handle outbound messaging reliably within platform constraints, I built dedicated send workers with Redis-backed rate limiting and retry logic, sustaining ~5 sends/sec per worker.
On the AI side, I engineered RAG-grounded reply generation using Gemini, embeddings, and pgvector — allowing the system to generate on-brand, contextually accurate responses grounded in each customer’s own knowledge base, alongside intent/sentiment classification and spam detection for incoming messages.
I also designed the comment-to-DM automation engine itself, which now supports 1,000+ configured automations with branching conversation logic and AI-agent handoff, backed by real-time state management for 10,000+ concurrent in-progress conversations. To make this configurable for non-technical users, I built a visual workflow engine and Next.js dashboard supporting branching conversation trees, multi-turn DM flows, follow/email gating, CRM-style lead management, and live analytics.
Beyond the core product, I worked on monetization and observability — integrating Dodo Payments for subscriptions and webhooks, building usage-based plan limits, tracked short-links, and analytics pipelines covering CTR, lead generation, and automation-level performance.
This role has given me end-to-end ownership across distributed systems, AI pipelines, and product-facing dashboards — a mix I’ve found genuinely energizing to work across.