AI @union_ai | Hosting AI Book Club | Teaching: MLOps, LLMs, GenAI, Computer Vision, and Robotics.
Joined June 2010
- Tweets 3,828
- Following 614
- Followers 1,637
- Likes 16,409
588 Photos and videos
Pinned Tweet
Mar 4
The Flyte 2.0 SDK is officially here: This release brings some really exciting local development features that you can right out of the box even with connecting to Kubernetes or Docker!
The features I've been loving to my local AI workflows:
- Terminal User Interface (live workflows and history)
- HTML Reports (tracks to workflow version
- Caching (skip rerunning a task by reading from cache)
- Retries (auto retry a task if it fails - super useful for API calls)
- And more
All this together gives you:
- great lightweight experiment tracking
- a lift in reliability
- a great interface for debugging large pipelines or AI agent runs
I can't wait to show you whats next in both Flyte and @union_ai
2
3
4
663
Jun 1
June's AI book club is Agentic Architectural Patterns for Building Multi-Agent Systems. From a commuity reccomendation.
1
103
Jun 1
I think I'll finally try streaming on X this month as well.
But if you want better reminder for the event you can RSVP on luma: luma.com/jzp46kbo
1
1
70
Jun 1
Chapters:
Part 1: Foundations and Core Agent Concepts
Chapter 1: GenAI in the Enterprise: Landscape, Maturity, and Agent Focus
Chapter 2: Agent-Ready LLMs: Selection, Deployment, and Adaptation
Chapter 3: The Spectrum of LLM Adaptation for Agents: RAG to Fine-tuning
Part 2: Agentic AI: Architecture and Design Patterns
Chapter 4: Agentic AI Architecture: Components and Interactions
Chapter 5: Multi-Agent Coordination Patterns
Chapter 6: Explainability and Compliance Agentic Patterns
Chapter 7: Robustness and Fault Tolerance Patterns
Chapter 8: Human-Agent Interaction Patterns
Chapter 9: Agent-Level Patterns
Chapter 10: System-Level Patterns for Production Readiness
Chapter 11: Advanced Adaptation: Building Agents That Learn
Part 3: Execution: Strategy, Use Cases, and The Future
Chapter 12: A Practical Roadmap: Implementing Agentic Patterns by Maturity Level
Chapter 13: Use Case: A Single Agent for Loan Processing
Chapter 14: Use Case: A Multi-Agent System for Loan Processing
Chapter 15: Agent Frameworks – Use Case: A Multi-Agent System for Loan Processing with CrewAI and LangGraph
Chapter 16: Conclusion: Charting Your Agentic AI Journey
52
May 19
When this month is over I'll have put on
- 4 AI Engineering workshops
- 4 AI engineer streams ( casual discussion and demos)
- 1 in person conference talk
Is this sustainable? We'll find out 😅
1
1
72
May 18
This Friday we'll be checking out @cognee_ ! It gives AI agents a shared, improving memory of your data, decisions, and workflows so they can recall, connect, and act with context.
Cognee is an open-source memory control plane for your Agents that lets you ingest data in any format or structure and continuously learns to provide the right context. It combines embeddings, graphs and cognitive science approaches to make your documents both searchable by meaning and connected by relationships as they change and evolve
4
3
281
May 13
Nothing like being on the "nice" airport wifi after conference wifi 😌
58
May 12
Wrapping up the embedded vision summit!
It's always incredible to see how people are pushing AI capabilities to the edge with a lot of constraints.
This was the most I've ever had people come up to me throughout the conference after my talk :)
1
59
May 11
Back in the Bay for embedded vision summit!
If you're going would love to meetup!
1
99
Apr 30
I'm excited to be back at Embedded Vision Summit again this year!
3
71
Apr 30
The Gemma4 26B A4B does an amazing job of provdiing frame by frame naration! In this case I'm also preserving the history of several frames as context so it knows over time how the scene is changing.
@NVIDIAAI DGX Spark Flyte 2 Devbox is becoming such a great AI lab combo 😎
1
5
217
Apr 20
This week we'll be building with Gemma 4!
Gemma 4 is Google DeepMind's latest open model family, built from Gemini 3 research and released under Apache 2.0. It ships in four sizes from Effective 2B up to 31B Dense, with native function calling, configurable thinking modes, and multimodal input across a 256K context window. Plenty of surface area for agentic workflows and fine-tuning experiments.
I've been testing it's visual understanding for scene comprehension and object detection. Very impressed so far, and I'm excited to test on the smaller versions to see how they do!
1
3
236
Apr 16
Using the Flyte sdk TUI for watching your autoresearch is cool!
You get this now with just pip install flyte[tui]>=2 btw!
1
3
95
Apr 16
38