Optimized GPU pipelines and built real-time, agentic AI microservices for video generation and debugging.

AI Engineering Intern

Role Overview

During my time at PGAGI, I engineered high-performance AI solutions, focusing on pipeline optimization, real-time communication, and agentic workflows.

High-Performance Computing & Video Pipelines

• Redesigned the short video generation pipeline running on H100 GPUs
• Implemented batch processing for 25-30 concurrent streams
• Slashed processing latency by 80%, reducing rendering time from 5 minutes to under 1 minute

Real-Time AI Microservices

• Architected a real-time interview microservice utilizing the Gemini Live API
• Integrated WebSockets to handle low-latency, bi-directional audio streaming
• Eliminated processing delays for a seamless conversational experience

Agentic AI Development

• Engineered an autonomous debugging AI agent for a Figma plugin
• Combined rule-based logic with an agentic pipeline to automatically identify and resolve MJML errors
• Developed automated lead generation scripts combining web scraping with personalized WhatsApp and Email outreach