Nathan Telles

About

Meta engineer who owned end-to-end production systems for AI-enabled smart glasses, spanning React/TypeScript apps through Node/Java microservices on AWS EKS with Kafka/Postgres. Built and productionized a real-time RAG pipeline (LangChain + OpenAI + Elasticsearch) with rigorous guardrails (shadow/canary, fallbacks, monitoring), delivering major improvements in latency (~35–40%), error reduction (~30%), and engagement (reported +40% DAU).

Key Strengths

• Owned and shipped end-to-end real-time AI response pipeline (frontend, backend, data, infra) on AWS EKS
• Scaled systems to thousands of concurrent requests with sub-second latency and near-100% uptime as usage tripled
• Latency optimization (profiling, Redis caching, async/parallel I/O, DB indexing) achieving ~35–40% latency reduction
• Productionized AI-native RAG capability (LangChain + OpenAI + Elasticsearch) with A/B testing and guardrails
• Reliability engineering for AI systems: fallback logic, confidence scoring, content filters, shadow testing, canary rollouts
• Legacy modernization leadership: monolith-to-microservices migration with compatibility layer, feature flags, shadow traffic, zero downtime
• Strong cross-functional execution with ML/AI research, backend, infra, and CPO/CTO stakeholders; translated ambiguous goals into measurable roadmaps
• Effective use of AI dev tools to accelerate architecture/refactoring (e.g., Kafka partitioning trade-offs, migration scripts) cutting effort ~40%

Experience

Education

Skills

Languages