Mohammad Zarei
As a Senior AI Engineer at Summer, I specialize in developing cutting-edge Agentic AI systems, most notably the "Sunny" platform, an advanced AI-driven assistant for real-estate market intelligence.
About
As a Senior AI Engineer at Summer, I specialize in developing cutting-edge Agentic AI systems, most notably the “Sunny” platform, an advanced AI-driven assistant for real-estate market intelligence. My work includes designing conversational workflows, analytical tools for exploring property listings, and ensuring scalable, reliable agent execution through robust production infrastructure and observability systems.
With a PhD in Engineering from the University of Waterloo, my expertise spans LLMs, retrieval-augmented generation, and production-grade machine learning pipelines. My career includes impactful roles in fintech, insurance, and computer vision, where I contributed to automating complex processes and enabling data-driven decision-making. I am committed to building scalable, intelligent AI solutions that address real-world challenges.
- 1 year's experience
- Joined December 2025
Industry experience
- Software & Internet
- Professional Services
More experts in Waterloo, Canada
Work experience· 5 roles
Developed Agentic platform 'Sunny', an AI-driven assistant for real-estate market intelligence. Designed conversational workflows and custom analytical tools enabling investors to explore and evaluate 1M+ property listings in real time. Engineered end-to-end observability and production infrastructure, integrating deep tracing, Logfire instrumentation, regression tests, thread-based conversation management, tool-call/prompt semantic caching, async I/O, and real-time streaming with FastAPI and Redis pub/sub, enabling reliable and scalable agent execution.
Implemented end-to-end Gen AI features/APIs powering AI-driven portfolio-analysis agents including personalized prompt library/recommendation/monitoring, embedding-based search, RAG-based commentary writer, web crawler, memory library, and planner. Utilized model routing, semantic caching, prompt optimization, input/output sanitation, secret management, and proper fallback strategies to improve system reliability.
Automated 3D scan quality assessment pipelines using multimodal Generative AI models, reducing manual review effort. Applied large vision-language models for diagnosing photogrammetry issues, and automated 3D scan quality assessment saving 10+ h/month.
Education
Certifications and awards· 7
Hire Mohammad Zarei today
Post a job and invite Mohammad, or contact them directly. No hidden commission.