Press, Talks & Writing
A chronological archive of third-party coverage, recorded talks and panels, interviews, and longer-form technical writing. Academic papers remain on the Publications page.
Last updated: August 2026
Press & Expert Commentary
-
Measuring What Matters in Large Language Model Performance
“We tend to think that we can identify reasoning when we see it. But there’s no universal consensus.”
-
Learning How Learning Works
“Training on additional data for a given language will only make the model more accurate in terms of predicting that next token.”
-
Can AI Make the Team?
“Think of LLMs as savants that can speak more intelligently about any topic under the sun.”
-
Scrutinizing LLM Reasoning Models
“If you can reward the model for being more concise, you can show a significant increase in accuracy across domains.”
-
The Real, Significant Threat of Shadow AI
“Employers are unfortunately not able to keep pace with the high demand for trying out generative AI models, tools, and apps.”
-
AI and Cybersecurity
Talamadupula argues that agents’ data access should be audited automatically after every system or configuration change.
-
Making Conversation a Robot’s Command
“Used with appropriate guardrails and sufficient context, this can significantly scale up the real world applicability of robotic applications.”
-
Safety Fears Raised Over Risks of ‘Penetrative AI’
“Hallucinations can be even more dangerous than in mere chatbots.”
-
Forget the Catastrophic Forgetting
“(It) is a significant problem for all machine learning models and systems.”
Public Talks, Panels & Interviews
-
Open Agent Specification: Enabling Cross-Framework Comparison of AI Agents
An introduction to Agent Spec, a framework-neutral declarative language for representing and comparing AI agents.
-
AAAI Presidential Panel: AI Agents
A discussion of how agents evolved from rule-based systems and multi-agent research into today’s LLM-powered systems.
-
Designing Intelligent Agents: The Role of Memory, Cognition, and Planning
A panel on cognitive architectures, memory models, and long-horizon planning for goal-driven agents.
-
Inside the Mind of an AI Agent: Frameworks, Planning, Memory and Tools
A panel examining the planning methods, memory systems, and tool integrations behind contemporary agent frameworks.
-
Can LLMs Really Understand?
An invited talk on what it means for language models to understand language and human conversation.
-
LLMs for Human Conversation: The History, the Nuances, and the Future
A technical talk on the history of conversational AI and the features that distinguish human conversation from generic text.
-
Building a Conversation-Focused LLM on Communication Data
A presentation on training a language model around the structure and nuances of communication data.
-
Nebula: A Conversation-Focused LLM Built on Communication Data
A walkthrough of Nebula and the research challenges involved in adding conversation and vision modalities to language models.
-
AI Aye Aye: A Discussion on Artificial Intelligence, Part 2
The second half of a wide-ranging discussion on AI adoption, applications, economics, language, and augmentation.
-
AI Aye Aye: A Discussion on Artificial Intelligence, Part 1
The first half of an expert conversation separating practical AI from common misconceptions and hype.
-
CLAI: Command Line AI from IBM
An introduction to IBM Research’s open-source platform for bringing AI assistance to the command line.
-
Project CLAI: Command Line AI
A concise overview of Project CLAI and its sense-act interface for AI capabilities in Bash.
-
AAAI-17 Talk on UbuntuWorld 1.0 LTS
A conference talk on a platform for automated problem solving and troubleshooting in Ubuntu.
-
Planning Challenges in Human-Robot Teaming: Ph.D. Defense
The doctoral defense covering automated planning challenges for robots acting in open, changing worlds alongside humans.
-
EasyChair as a Pedagogical Tool: Engaging Graduate Students in Reviewing
A talk on using a conference-review platform to teach graduate students how peer review works.
Technical Writing
-
Metering Is Coming
Why enterprise AI is moving from adoption-at-any-cost toward outcome-per-token economics and metered cognition.
-
Concise Reasoning via Reinforcement Learning: Interactive Report
An interactive explanation of the evidence and training dynamics behind concise reasoning in language models.
-
Can Conversational Feature Transfer in LLMs Help Detect Deception?
An empirical look at whether features learned from multimodal conversation data transfer to deception detection in text.
-
A Guide to Building an LLM from Scratch
A practical tour from use-case definition and architecture through data curation, training, and evaluation.
-
A Guide to Building an End-to-End Speech Recognition Model
A guide to the data, architecture, training, and evaluation stages of end-to-end automatic speech recognition.
-
Emotional Intelligence in LLMs: Evaluating Nebula on EQ-Bench and Judgemark
An evaluation of how a conversation-focused model recognizes and judges emotional nuance.
-
A Guide to Transformer Architecture
A component-by-component explanation of the transformer architecture behind modern language models.
-
A Guide to Overfitting and Underfitting
How bias, variance, data, and model complexity produce underfitting or overfitting—and how to diagnose them.
-
A Guide to AI Agents
An overview of agent architectures, benefits, applications, risks, and likely directions.
-
Are Human Conversations Special? A Language Model Perspective
Evidence that human conversations demand broader contextual attention than many other language domains.
-
A Guide to LLM Inference Performance Monitoring
Metrics and operational practices for tracking language-model quality, latency, reliability, and cost at inference time.
-
A Guide to LLM Hyperparameters
How training and inference hyperparameters shape model learning, variability, and output behavior.
-
A Guide to Quantization in LLMs
How lower-precision representations reduce model size and inference cost, including the resulting tradeoffs.
-
A Guide to Comparing Different LLM Chaining Frameworks
A comparison framework for selecting orchestration libraries for multi-step language-model applications.
-
Building a RAG-Based Conversational Application with Nebula LLM
A high-level implementation guide combining Nebula, Milvus, embeddings, and LangChain for conversational RAG.
-
Guide to Fine-Tuning Techniques for LLMs
A guide to full fine-tuning, transfer learning, RLHF, PEFT, LoRA, DPO, and their operational tradeoffs.
-
A Comprehensive Guide to Vector Databases
An explanation of embeddings, vector search, indexing, and their role in generative AI applications.
-
An In-Depth Guide to Benchmarking LLMs
A survey of benchmark design and the major capabilities, datasets, and metrics used to evaluate LLMs.
-
NeurIPS 2023: A Technical Summary
A field report on prominent research themes, papers, and talks from NeurIPS 2023.
-
Guide to Context in LLMs
How context windows, retrieval, prompting, and memory influence language-model responses.
-
LLM Frontiers: Multilingual Conversations
Why multilingual conversational systems need more than straightforward translation of text and prompts.
-
Understanding Spoken Language
The technical shortcomings of text-oriented language models when interpreting multimodal human conversation.