Ganesh Subramanya

I’m an AI Engineer with 1+ years of production experience designing and deploying production AI systems across real-time Voice AI, agentic RAG, computer vision, and enterprise workflow automation. I focus on shipping end-to-end solutions—from model integration and backend engineering to observability, deployment, and performance optimization.

At my current role, I engineered a real-time agentic Voice AI platform with streaming STT/TTS, Silero VAD, hybrid retrieval (Qdrant + BM25 + reranking), and interrupt-driven orchestration to reach 1.5–2.5s end-to-end conversational latency. I also built an enterprise AI assistant that combines RAG, tool calling, and workflow automation, using modular retrieval architecture to reduce manual knowledge lookup.

I’ve designed agentic systems for knowledge and evaluation: a markdown-based enterprise knowledge platform with semantic retrieval, versioned documentation, human-approved memory persistence, and a 6-step reasoning pipeline to improve response consistency. I also built an AI-powered candidate assessment platform that combines voice interview analysis, resume intelligence, and LLM reasoning, plus fraud scoring and anomaly detection pipelines that align closely with recruiter decisions.

On the computer-vision side, I architected an AI-powered candidate integrity monitoring platform using YOLO, MediaPipe, OpenCV, 6DRepNet, and real-time backend services, achieving sub-200ms inference latency and strong violation detection performance. Alongside this, I develop tools and products (e.g., agentic navigation and adaptive note-taking) to validate ideas quickly and turn them into usable, accessible experiences.