by Serj Smorodinsky
Talks
Presentations on evaluation-first AI engineering, DSPy, and shipping LLM systems you can trust.
Serverless Toronto
Serj Smorodinsky & Brett Kennedy · Co-authors of Building LLM Applications with DSPy (Manning)
Why evaluation is the bottleneck when coding agents write the code, and how DSPy turns data, metrics, and optimizers into short programs you can review. Walks through an ATIS intent-classification example from baseline to optimized, deployed system.
Nebius Academy
Serj Smorodinsky · NLP Research Team Lead at Loris.ai, Nebius Academy lecturer
How AI agents make decisions and where they fall short. Covers best practices for building agentic systems and how their design connects to human psychology.
Nebius Academy · Y-DATA Agentic Systems elective
Serj Smorodinsky · NLP Research Team Lead at Loris.ai
A standalone lecture from the Y-DATA agentic systems elective covering the tools behind modern agents: deep research agents, browser agents, and the Model Context Protocol (MCP).