
Upcoming AI Events in the Bay Area, April to May 2026: DataPhoenix Events Digest
Dmytro Spodarets·Apr 14, 2026
1:01:41
Dmytro Spodarets·Sep 18, 2026
1:02:51Tech Talks
Yogiraj Awati·Sep 17, 2026
1:17:55Tech Talks
Santosh Kumar Radha·Sep 17, 2026
48:04Tech Talks
Fabrizio Dimino·Sep 17, 2026
1:05:16Tech Talks
Rome Thorstenson·Sep 17, 2026
1:00:23Tech Talks
55:47Tech Talks
1:18:26Tech Talks
Reduce GPU memory use and training costs with 10 practical techniques, from LoRA and sequence packing to sharding and data selection.
Dmytro Spodarets·Sep 14, 2026
These 10 MLOps trends are worth watching in 2026 for their potential to improve agent reliability, evaluation, context management, inference efficiency, and deployment.
Dmytro Spodarets·Jan 5, 2026
Robutler is open-sourcing WebAgents - the framework for building connected AI agents that can discover and delegate to each other in real-time. WebAgents enables agents to dynamically extend their capabilities by discovering and orchestrating agents on demand.
Three lists, one email. Pick what fits — unsubscribe with one click.

Dmytro Spodarets·Apr 14, 2026




Dmytro Spodarets·Mar 14, 2025
Show your AI or data project in a live demo, or share your technical experience in a deep-dive talk. We’re opening submissions for both Data Phoenix shows this fall. Explore the formats, selection deadlines, and ways to participate.
Dmytro Spodarets·Sep 22, 2026
Lasso Security found that SynthID-Text shifted some tool calls and refusals in open-weight model tests, with the effect varying by model and watermark key.

Strands released an Apache-2.0 agent harness for Python and TypeScript, pairing built-in tools and context controls with model and container deployment options.

Databricks released Consort, a community-supported agent framework that freezes specifications, separates development roles and tests changes on isolated Lakebase branches before they pass human approval gates.

Databricks says Proteus keeps agent-written GPU kernel proposals separate from controlled validation, repeated timing and scoped context, while reporting faster shape-specific kernels for parts of Qwen3.5 122B.

OpenAI announced a nine-member mathematics advisory group. Its unpaid researchers may publish unsolicited advice but cannot make decisions for the company.

SpaceXAI has released Grok 4.7 through its API, Grok Build and Cursor, targeting coding and extended knowledge tasks with a 500,000-token context window.

AWS has detailed how customers can use xAI’s Grok 4.6 through Amazon Bedrock, including two endpoint options and three supported interfaces.