Completed
– UX for Agents: Why text boxes alone aren't enough
Class Central Classrooms beta
YouTube videos curated by Class Central.
Classroom Contents
Beyond Copilots - How LinkedIn Scales Multi-Agent Systems
Automatically move to the next video in the Classroom when playback concludes
- 1 – Evolution of Generative AI at LinkedIn: From "Coach" to "Agent"
- 2 – The Early Days: Simple prompt-in/string-out products
- 3 – Moving to Prompt Chains: Handling memory and online inference
- 4 – The "Agent Era": Introducing prompt graphs and task automation
- 5 – Deep Dive: The LinkedIn Hiring Assistant problem space
- 6 – Why natural language interfaces beat 40+ search filters
- 7 – Scaling bottlenecks in single LLM block architectures
- 8 – Modular Design: Moving to a Manager/Interpreter pattern
- 9 – Transitioning from LLM blocks to hierarchical sub-agents
- 10 – The Supervisor Pattern: Coordinating specialized agent skills
- 11 – Parallel development and independent quality evaluation
- 12 – Model Selection: When to use GPT-4o vs. fine-tuned small models
- 13 – Domain Adaptation: Training models on the LinkedIn Economic Graph
- 14 – The LinkedIn Agent Platform: Standardizing prompts and namespaces
- 15 – LLM Inference Abstractions: Managing quotas and GPU limits
- 16 – Scaling non-deterministic workloads with a messaging platform
- 17 – Memory Management: Working memory vs. long-term collective memory
- 18 – Building a Skill Registry and why it predated MCP
- 19 – Observability challenges in asynchronous agentic systems
- 20 – Lessons Learned: When to use procedural code instead of an LLM
- 21 – The Model Customization Pyramid: RAG vs. Fine-tuning
- 22 – UX for Agents: Why text boxes alone aren't enough
- 23 – Q&A: Managing security and service principles in a skill registry
