At Complya, I develop domain-specific AI calling agents and production LLM solutions, with a focus on low latency and cost-efficient inference. I built end-to-end RAG systems with OpenAI, Claude, and Vertex AI, and developed real-time conversational AI using FastAPI and WebSockets.
At Estela, I trained computer vision models for real-time detection and developed ML pipelines and model-serving APIs. I also worked on speech systems, NLP, and model optimization, including improving throughput from 5 FPS to 33 FPS.

