Gregory Chin

Software Engineer at Dayta Al

•Built real-time IoT data pipelines for the Cyclops retail intelligence platform — ingesting RTSP camera streams, processing 150k+ daily videos, and integrating POS and sensor data for foot-traffic heatmaps, demographic analytics, and peak-hour alerts. •Developed CV-powered demographic tracking (age/gender via OpenCV models) and anonymized RTSP stream processing with secure role-based dashboards for retail and mall operations teams. •Created script-to-video pipelines transforming raw store camera data into narrated shopper-behavior reports using FFmpeg and TTS — eliminating manual video editing for marketing teams entirely. •Implemented real-time traffic preview and cloud-rendered BI report pipelines (AWS MediaConvert + Three.js) that reduced analytics delivery time by ~60% for global retail clients. •Built Ruby on Rails REST APIs and background job workers (Sidekiq) to power the Cyclops dashboard's data ingestion and alert-delivery layer, enabling reliable multi-tenant data routing and configurable threshold-based notifications for retail operations teams. •Maintained CI/CD pipelines for reliable multi-tenant deployments; integrated real-time IoT event alerts with POS systems to support staffing decisions, campaign evaluation, and tenant mix optimization. •Designed and deployed a real-time video analytics system using Python, TypeScript, Node.js, and React, integrated with AWS and Firebase for scalable authentication and media processing, implemented WebSockets and REST APIs to handle RTSP streams from IoT sensors, and established automated CI/CD pipelines that improved system reliability and released updates faster.

Senior Software Engineer at Self Employed

•Built and launched voice AI agents that reduced response time by 30-40%, earned a 4.8/5 satisfaction score, and improved real-time interactions for healthcare and SaaS clients. •Built and deployed multi-stage RAG pipelines using LangChain, LangGraph, and Pinecone/Qdrant, improving information relevance and reducing query latency for clinical, insurance, and knowledge retrieval tasks. •Built multi-agent AI systems with LangGraph: orchestrator + domain agents with defined tool sets, shared memory, and output guardrails — automating healthcare intake, insurance pre-auth, CRM/EMS data entry, and appointment scheduling workflows; reduced manual processing time by ~50%. •Leveraged AWS Bedrock and Bedrock Data Automation to build document extraction, summarization, and routing pipelines, and used LangSmith for tracing, evaluation, and iterative prompt optimization in production; this increased extraction accuracy and lowered manual review effort. •Integrated Azure AI Foundry and Azure OpenAI to deploy and evaluate enterprise LLM solutions, using Azure AI Search for hybrid retrieval (vector + keyword) over structured and unstructured enterprise data sources; improved retrieval precision and enabled scalable, secure deployment across Azure-hosted environments. •Developed AI-powered outbound/inbound lead generation agents integrated with CRMs via gRPC and webhook APIs — increasing qualified lead volume by 30-40% and cutting manual calling effort by ~50% for SMB clients. •Built fintech and marketing automation pipelines (email/DM sequencing, LLM personalization, follow-up logic) using n8n and FastAPI — lifting reply rates by 20-25% and compressing campaign setup from days to hours. •Delivered full system architecture per engagement: requirements, data modeling, gRPC/REST API design, Docker containerization, and AWS deployment (Lambda, API Gateway) with structured observability — latency, error rates, and prompt audit logs via CloudWatch and LangSmith. •Developed Java Spring Boot microservices and C# ASP.NET Core APIs for enterprise client integrations, leveraging .NET's dependency injection and middleware pipeline to expose secure, versioned REST endpoints consumed by web and mobile frontends. •Authored performance-critical Rust modules for low-latency data transformation and stream processing tasks where throughput and memory safety were essential; integrated compiled Rust components via FFI into existing Python and Node.js service boundaries. •Built Ruby on Rails backends for rapid-prototyping engagements, using ActiveRecord, background jobs (Sidekiq), and RESTful conventions to deliver functional MVPs quickly for early-stage SaaS clients. •Leveraged Python, Go, and TypeScript with FastAPI, NestJS, PostgreSQL, Redis, AWS Bedrock, Pinecone, Qdrant, LangChain, LangGraph, LangSmith, LiveKit, Twilio, gRPC, Kafka, n8n, and Docker to build scalable AI-enabled services that powered multiple fintech and marketing automation projects, increasing system throughput

Senior Software Engineer at Butterflies Al

•Led engineering for a 7-person team building an AI social network, designing and implementing LLM backends, agent pipelines, real-time event systems, and cross-platform APIs (web, iOS, React Native) on AWS Bedrock and LangGraph, supporting millions of daily users. •Architected autonomous AI agent pipelines for thousands of concurrent characters, handling multi-turn memory, persona consistency, safety filtering, and prompt orchestration, and added LLM-as-judge evaluation that improved output quality and reduced moderation incidents. •Built generative character creation services using ComfyUI + Stable Diffusion + custom LoRA training pipelines, enabling AI persona creation in minutes and driving thousands of new characters per week. •Designed event-driven microservices (message queues, priority workers, Pub/Sub patterns) for image generation, feed updates, and notification delivery — using batch processing and queue prioritization to sustain sub-second feed latency at millions of AI-generated events per day. •Implemented scalable backend services in Node.js and Supabase (row-level security, auth, rate limiting) on AWS; shipped features that grew AIgenerated content 50%+ quarter-over-quarter and measurably improved user retention. •Standardized APIs, designed data models, and set up observability using MySQL and AWS Bedrock; mentored junior engineers through design docs, code reviews, and pair programming on LLM-driven backend challenges, which accelerated onboarding and reduced integration bugs. •Developed and deployed end-to-end features with Python, Node.js, NestJS, Supabase, PostgreSQL, React Native, and AWS services, leveraging LangChain, OpenAI, and Stable Diffusion to add script-to-video generation, thereby expanding the product's AI capabilities and boosting user engagement. •Wrote Java-based data ingestion services and C# .NET utility tooling to support internal analytics pipelines and cross-team integrations, maintaining consistent coding standards across a polyglot engineering org.