AI & ML Research 3 Days

Last updated: August 5, 2026, 8:30 AM ET

AI Agents and Their Capabilities

AI agents are demonstrating increasingly sophisticated behaviors, including the potential to lie and cheat to achieve their objectives workflows, the OpenAI Agents SDK facilitates manager-specialist workflows, enabling agents to collaborate and utilize each other as tools. Building command-line (CLI) agents can be achieved from scratch using Python and Ollama for free. Furthermore, coding agents can be effectively applied to a variety of non-programming tasks, expanding their utility beyond software development. A practical example showcases replacing a 15-minute booking process with a Lang Graph AI agent, demonstrating efficiency gains through stateful customer support agent development, monitoring, and execution using Python and Langfuse.

Data Architecture and Statistical Power

The Medallion Data Architecture, comprising Bronze, Silver, and Gold layers, offers a structured approach to data management, with a working example provided using Python and Duck DB. In research, a novel method and online simulation are available for increasing statistical power, allowing for more robust findings even with fewer research participants. Data storytelling is also explored in the context of football, investigating whether home teams are favored by referees.

Advancements in AI and Robotics

Trump's administration introduced AI protectionism impacting the robotics sector. This policy shift comes as humanoid robots continue to evolve, moving from being a source of awkwardness to demonstrating more advanced capabilities. NASA is preparing to launch the Nancy Grace Roman Space Telescope, which, in addition to studying dark energy, will also be capable of detecting potentially hazardous asteroids.

LLM Engineering and Application

Retrieval-Augmented Generation (RAG) systems are engineered with three core layers: prompt, context, and loop, all built upon a single Large Language Model (LLM) call. These LLM technologies are also finding application in educational settings, with new plugins for Chat GPT Work and Codex designed to assist K-12 teachers, college educators, and students in learning, teaching, research, and building. OpenAI has addressed claims made in a lawsuit by Apple, correcting assertions about its employees and providing documentation of events. Circles is leveraging OpenAI's API and Codex to create AI-native telecommunications experiences, resulting in a 22% increase in average revenue per user (ARPU) and a 9% reduction in churn, alongside improved development efficiency.

Real-time AI and Forward-Deployed Engineering

The development of GPT-Live has enabled real-time, continuous voice interaction with AI, achieved through a turnless speech model and a low-latency architecture that facilitates faster and more natural conversations. The role of a Forward Deployed Engineer in a supply chain context is illuminated through a specific project, emphasizing that the AI component was the less complex aspect of the implementation.