I designed a 51-node n8n pipeline for AI video production and multi-platform publishing. It connects script generation, AI image and video generation, voice synthesis, video composition, and publishing APIs.
I integrated OpenAI, ElevenLabs, Creatomate, and other external services, building structured data flows and handling JSON transformations, authentication, and workflow dependencies. I also made the stages modular so they could be tested and maintained independently.
For OmniShorts, I built the foundation of an automated video-processing platform using PostgreSQL, Redis/BullMQ, Supabase storage, yt-dlp, and FFmpeg. I implemented asynchronous job processing with retry and backoff, persistent job-state tracking, and real-time progress events.
I also built DOOMBOT, a multi-agent LLM system for structured trading research, with knowledge retrieval, tool calling, and a deterministic 90-case reasoning dataset. As a Data Analyst Intern at Bluestock, I used Python, Pandas, NumPy, and SQL to analyze data and build eight dashboards.
