Tuesday, February 09, 2027
9:00-9:45am
Tuesday - 2/09/27
|
|
10:00-10:45am
Tuesday - 2/09/27
|
AI Developer (Room 251A)
Engineering AI-Ready Compute: Building Reliable, High-Performance Platforms at Scale
AID-02 This session explores how quality-first validation and system engineering improve the performance, stability, and readiness of computing platforms for AI workloads. Drawing from high-performance computing (HPC) and CPU/GPU validation environments, it covers regression and stress-testing strategies alongside advanced debugging across hardware, firmware, OS drivers, and virtualization layers.
Register me 
|
11:00-11:45am
Tuesday - 2/09/27
|
AI Developer (Room 251A)
The Productivity Reality Check: What AI Coding Tools Actually Deliver in Regulated Industries
AID-03 AI coding assistants promise rapid development, but in regulated fintech, speed must survive audit trails and strict compliance. This practitioner-level talk evaluates daily use of AI dev tools across code review, test generation, documentation, and debugging. It contrasts time saved against unexpected friction, hallucination risks, and code maintenance overhead.
Register me
|
12:00-12:30pm
Tuesday - 2/09/27
|
|
1:30-2:15pm
Tuesday - 2/09/27
|
AI Developer (Room 251A)
Long-running agent tasks often fail when process memory is wiped by container restarts, autoscaling, or redeployments. This session demonstrates how to externalize agent state by containerizing an agent loop on Red Hat OpenShift, moving transcripts out of RAM into Redis, and serving inference via Red Hat OpenShift AI (vLLM). Includes a live pod-kill demo showing mid-task execution recovery.
Register me
|
2:30-3:15pm
Tuesday - 2/09/27
|
AI Developer (Room 251A)
Traditional software testing validates data structure, not semantic meaning—leaving systems vulnerable to silent distribution shifts where schema checks pass but underlying data is corrupted. This engineering deep dive details how silent corruption occurs in production, why AI-generated data amplifies the risk, and how to build lightweight detection layers that flag semantic anomalies before they reach production models.
Register me
|
3:30-4:00pm
Tuesday - 2/09/27
|
|
Wednesday, February 10, 2027
9:00-9:45am
Wednesday - 2/10/27
|
|
10:00-10:45am
Wednesday - 2/10/27
|
AI Developer (Room 251A)
Managing high-speed AI agent execution with traditional human sprint cycles creates operational bottlenecks. This talk outlines the AI-Driven Development Lifecycle (AI-DLC), replacing sprints with short execution "Bolts." Specialized agents handle tasks across Inception, Construction, and Operations, using lifecycle checkpoints as orchestration boundaries to maintain human oversight.
Register me
|
1:30-2:15pm
Wednesday - 2/10/27
|
AI Developer (Room 251A)
Choosing an Agentic Framework: LangGraph, AutoGen, CrewAI, LlamaIndex Agents, and the New Wave
AID-10 Switching agent frameworks mid-production is as complex as a database migration. This vendor-neutral session evaluates leading choices—LangGraph, AutoGen, CrewAI, and LlamaIndex Agents—across critical production dimensions: state and memory management, tool integration models (MCP vs. proprietary), production durability, multi-agent coordination topologies, offline evals, and post-ship migration costs.
Register me
|
2:30-3:15pm
Wednesday - 2/10/27
|
AI Developer (Room 251A)
Building High-Performance Java & Spring Boot Applications for Peak Traffic Events
AID-11 Peak events like holiday sales put extreme demands on enterprise infrastructure. Drawing from mission-critical platforms at retail scale, this session covers proven architectural patterns, performance tuning, and operational practices to survive sudden traffic spikes. Topics include microservice design, JVM tuning, connection pooling, caching, circuit breakers, and cloud scaling on GCP.
Register me
|
Thursday, February 11, 2027
9:00-9:45am
Thursday - 2/11/27
|
|
10:00-10:45am
Thursday - 2/11/27
|
AI Developer (Room 251A)
Adapting LLMs mid-session usually means bloating context windows and paying a heavy memory tax. Emerging research shifts adaptation into the inference loop itself via Test-Time Training (TTT), Test-Time Adaptive Optimization (TAO), and d-mem. Featuring a live local demonstration on the MLX engine, watch an agent adapt on the fly across Apple Silicon's unified memory during an active incident.
Register me
|
11:00-11:45am
Thursday - 2/11/27
|
AI Developer (Room 251A)
AI test agents mark a shift from deterministic scripts to adaptive quality engineering systems capable of interpreting natural language and navigating changing UIs. This session presents a research-grounded framework for deploying probabilistic test agents alongside existing automation frameworks. Learn how to establish accuracy thresholds, confidence-based escalations, and fallback strategies that maintain test resilience.
Register me
|
