Talks

Some lectures and invited talks I have delivered publicly, with recordings available online. The first two lectures were first delivered in the CISC877: Software Engineering and Foundation Models graduate course at Queen's University, Kingston, Canada, and variants were later presented at AIware Bootcamps in Toronto and Ottawa for a broader audience.


Agentic Architectures and Workflows

CISC877 @ Queen's AIware Bootcamp Lecture 90 mins

A deep dive into how foundation model agents are built and orchestrated, from first principles to multi-agent systems in software engineering.

Topics covered:

  • What is an Agent? (e.g., Perception, Brain, Action, Environment)
  • Different types of agentic memory
  • How does an Agent use tools? (Toolformer, ReAct)
  • How to enable an Agent to plan and reason? — Introduction to the theory of mind
  • Cognitive architectures — Multi-layered abstractions (recursion, multi vs single agent architectures)
  • Agent patterns — Chains, routers, graphs
  • Control mechanisms — Static vs dynamic orchestration
  • Patterns of multi-agent collaboration
  • Self-reflection and multi-agent hacks
  • Agents in software engineering

RAG Engineering

CISC877 @ Queen's AIware Bootcamp Lecture 90 mins

From the fundamentals of retrieval-augmented generation to the advanced techniques needed for enterprise-grade, production-ready RAG systems.

Topics covered:

  • What is RAG? — An overview of indexing, retrieval, and generation
  • An overview of query translation approaches — Multi-query, RAG-fusion, Least-to-most, step-back prompting, HyDE
  • Query routing strategies — Logical routing and semantic routing
  • Query construction strategies
  • Advanced indexing techniques for effective RAG — Multi-representation indexing, RAPTOR, ColBERT
  • Graph RAGs
  • RAG re-rankers
  • Large context challenges: Needle in a haystack, counting stars, data movement and caching/pinning of context in the cloud
  • Context vs built-in prioritization
  • RAG in software engineering research
  • A reference architecture for enterprise-grade production-ready RAGware

Beyond Vibe Coding: Leveraging Foundation Models to Infer Intent

RAISE'25 @ ICSE'25 AIware Bootcamp Europe Invited Talk 30 mins

First delivered at the RAISE 2025: Requirements engineering for AI-powered SoftwarE workshop, co-located with ICSE'25 in Ottawa. A variant of this talk was later delivered at AIware Bootcamp — Europe 2025 in Amsterdam, Netherlands, co-located with OSS Europe 2025.

Topics covered:

  • Why requirements engineering is the bottleneck — Aligning software with true user needs starts with understanding intent
  • From vibe coding to real engineering — Critiquing the throwaway culture of prompting and patching
  • What are intents, really? — Distinguishing goals, requirements, and the hidden rationale that underpins them
  • Inferring intents with foundation models — Using Theory of Mind, iterative refinement, and dialogue context to bridge ambiguity
  • Conversational interfaces for requirement elicitation — Proactively clarify requirements, resolve contradictions, and avoid premature solutions
  • Evaluating intent quality — Rubrics for accuracy, relevance, completeness, realizability, and collaboration support
  • Real-world gains over baselines — Evidence that FM-inferred requirements deliver better grounding and satisfaction
  • Known challenges — Tackling sycophancy, hallucination, and compute cost
  • What's next for requirements engineering — Reasoning-augmented models, multi-modal inputs, and non-chat interfaces
  • Call to action — Foundation models shouldn't just write code; they must understand why the code exists