About the role
At Twilio, the Machine Learning Engineer will join the Conversation Memory team, which develops the memory layer for Twilio's Conversational AI platform. This team focuses on extracting relevant information from vast amounts of conversation data, associating it correctly with customers, and enabling rapid retrieval to support customer interactions. The role supports Twilio's remote-first culture and global impact in communications technology.
The engineer will build and maintain pipelines that convert raw conversation data into structured, searchable memory, tuning large language model prompts and schemas. They will also manage the retrieval systems using embeddings and vector search, and develop evaluation frameworks to ensure memory accuracy. Daily tasks include diagnosing production issues and collaborating with product and ML teams to improve system performance and quality benchmarks.
Candidates should have experience building backend services and a strong curiosity for system behavior in production. Skills in observability, evaluation systems, and AI-assisted development tools are important. Preferred qualifications include knowledge of Go, Python, event-streaming systems like Kafka, and experience with LLM-backed products or machine learning. The position is remote within the US, excluding certain California areas, with occasional travel for team meetings.