Joshua Sommers

I am a senior software engineer with 8+ years of experience building production-grade AI and LLM systems, currently focused on RAG architectures, multi-agent orchestration, and LLM evaluation/guardrails. At Fundrise, I architected an embedded investor copilot used by thousands of active users, built on a multi-agent orchestration layer with a production RAG pipeline, and drove hallucination rates down to low single digits through strict guardrails and compliance validation. I also led an AI decisioning platform combining uplift/churn models with RL-driven policy selection, handling 100-300 decisions/sec at sub-100ms latency and delivering up to 4x lift in re-investment for dormant user cohorts. My core stack is Python(FastAPI), with deep experience across LangChain/LangGraph, PGVector, TensorFlow/PyTorch, and cloud infrastructure spanning AWS, GCP, and Azure. I've spent my career at the intersection of ML/NLP and full-stack engineering-from intelligent document processing to sales intelligence platforms to predictive maintenance systems-always shipping models that go from prototype to reliable, monitored production services. I'm looking for a senior or staff-level remote role where I can own AI/LLM systems end-to-end, ideally at a product company building applied AI at scale.