
This Week in #MachineLearning & #AI (podcast) brings you the most interesting and important stories from the world of #ML and artificial intelligence.
Joined May 2016
- Tweets8.6K
- Following1.6K
- Followers14.2K
- Likes2.8K
Pinned Tweet
An AI agent can make many model calls to finish a single job. Some need frontier reasoning. Others can run on an open-weight model, and some involve data that has to stay in-house. How do you decide which inference to buy and which to run yourself?
Sadeepa Wijesekara, who leads engineering for @HPE_AI Private Cloud AI, joins Sam Charrington to discuss how enterprises make those decisions. They get into what it takes to see the cost of individual calls, how teams route work between hosted and local models, and when running your own GPUs makes economic sense. Sadeepa shares how HPE routes coding requests through a single internal endpoint that can send work to a frontier provider or serve it locally.
Full show notes: twimlai.com/go/785.
📖 CHAPTERS ===============================
00:00 - Intro
01:42 - Why agent workloads are driving up token bills
04:27 - Measuring agent value by business outcome
06:37 - Token budgets and smarter model choices
09:14 - Rule-based versus automated model routing
11:07 - Making token spend visible at the agent and call level
16:06 - Workload placement across hosted and local models
18:11 - Standing up on-prem inference
20:00 - GPU selection, scheduling, and utilization
24:38 - Building your own infrastructure versus buying a platform
26:39 - Layered security for autonomous agents
30:48 - OpenShell, access controls, and hardware attestation
33:25 - Enterprise readiness and agent compliance
37:02 - Inside HPE’s hybrid AI factory
40:27 - Sizing and scaling internal inference capacity
42:15 - Comparing inference costs and protecting sensitive data
48:55 - Customer deployments: sovereignty, manufacturing, and shared inference
54:02 - Moving from token-cost controls to measurable ROI
56:34 - What’s next for agent security
2
361
How do you keep AI spending under control as agents consume more tokens, while giving teams room to experiment and keeping the quality of the work high?
In this TWIML Briefing Room episode, Sam Charrington talks with Dan Waibel, @HPE_AI 's Global Chief Data and AI Officer, about how HPE is managing that trade-off across its own business. Waibel explains how HPE uses token usage data and quality evaluations to decide which models handle which jobs, where running inference in-house makes financial sense, and when it’s worth paying for cloud models. He also shares how HPE gives teams a fast path to experiment and measures whether the AI projects it scales are improving business performance.
📖 CHAPTERS
===============================
00:00 - Intro
01:06 - Tokenomics and the growth of agent-driven token usage
05:28 - Tracking token spend without encouraging waste
07:33 - From individual productivity to AI-first workflows
09:46 - The economics of hybrid AI and HPE’s on-prem target
12:11 - Security, privacy, and protecting company know-how
14:00 - How always-on agents change the infrastructure equation
15:41 - Measuring AI investments against business KPIs
17:52 - When to consider an AI factory—and how to govern it
20:30 - Using observability and evals to improve model routing
Listen to the companion episode: Buying Tokens vs Building an AI Factory with Sadeepa Wijesekara: twimlai.com/go/785
Stefano Corazza highlights the importance of the agent harness, explaining that it was one of the hardest parts of developing Canva Brush and required multiple iterations. Canva’s approach involved making the harness thinner, simpler, and more flexible, giving the underlying LLM greater freedom to use tools and operate on designs. Agent performance depends not only on the capabilities of the model but also on the systems and tools surrounding it.
Watch the full conversation on the TWIML AI Podcast: youtu.be/heE80uxOmus
1
224
For creative professionals, changing the position of an object, adjusting its size, or modifying one element shouldn’t require regenerating the entire image and hoping everything else stays consistent. Stefano Corazza explains Canva’s approach to making AI-generated images more editable through structured elements and Magic Layers. The bigger takeaway? Consistency and control are just as important as image quality when bringing generative AI into professional design workflows.
Watch the full conversation on the TWIML AI Podcast: youtu.be/heE80uxOmus
A perfect first generation is a poor assumption to build a creative tool around. Stefano Corazza, head of AI research at Canva, calls out the Silicon Valley bias and explains why creative AI workflows need room for revision and need to be built around iteration rather than single-shot generation.
Watch the full conversation on the TWIML AI Podcast: youtu.be/heE80uxOmus
287
The future of AI inference is multi-architecture, giving teams many more hardware options than just GPUs. Custom accelerators from different vendors are already finding their niches, sometimes within the same workload. The question is whether a common software stack can make those choices practical while taking advantage of what each chip does best.
In this episode, recorded at Qualcomm’s Snapdragon Summit 2026, @clattner_llvm, EVP of Advanced AI Software and Platforms at @Qualcomm and co-founder of @Modular, joins us to discuss what it takes to build for a multivendor AI ecosystem. We explore how Mojo and MAX share serving logic across chip architectures while tuning performance-critical kernels for each, why Qualcomm open-sourced Mojo after acquiring Modular, and how that approach could extend from data centers to laptops and phones. Chris also shares where his team uses coding agents to optimize kernels and why engineers still own architectural decisions.
📖 CHAPTERS
===============================
00:00 - Intro and Chris’s path from Swift to Modular
01:58 - Why AI inference needs different chips
03:21 - Sharing the stack without sacrificing chip performance
05:47 - Why Modular joined Qualcomm and open-sourced Mojo
08:10 - Bringing the AI stack to laptops, phones, and beyond
09:42 - Mojo’s future as a general-purpose language
12:13 - AI progress, hype, and what coding agents don’t replace
14:30 - How Chris uses AI coding tools
15:27 - Letting agents optimize kernels while humans own architecture
16:15 - Where agentic coding goes next
2
521
An hour to go until our Generative AI study group starts! Exchange hardware acceleration techniques, dig into hybrid attention models, explore recursive self-improvement, and more. See you there!
2
2
619
Just one day until our next Generative AI study group! Explore agent sandboxing, share hyperparameters and tuning strategies, discuss model quantization approaches, and more. Sign up at twimlai.com/community/ and we’ll see you tomorrow at 8 am PT.
1
351
Our Generative AI study group centers around sharing insights and experiences with AI platforms, comparing newly released models and features, exploring developer tools and products, and more! Anyone interested in joining can register at twimlai.com/community/ and join us every Friday at 8 am PT.
1
269
In this episode, Stefano Corazza, head of AI research at @canva, joins us to explore how his team is building custom models and agents around iterative design workflows. We discuss the trade-offs between training models from scratch and adapting open weights models, and how proprietary design data supports specialization. Stefano explains how Craft separates layout planning from pixel generation and how Brush uses design tools and visual feedback to create and revise compositions. We examine the challenges of evaluating design quality beyond a single generation, including the use of vision language models as judges alongside human evaluation. We also discuss lessons from simplifying agent harnesses, why faster inference enables more iterations, and how model ownership creates opportunities to optimize serving costs. Finally, we explore how APIs and MCP are expanding access to these capabilities and what changes when software companies build tools for agents as well as people.
🗒️Full show notes: twimlai.com/go/784.
📖 CHAPTERS
===============================
00:00 - Introduction
01:19 - Canva AI Research, Craft, and Brush
04:25 - Why Canva Builds Custom Models
09:43 - Canva Craft Layout-to-Pixels Architecture
12:27 - Canva Magic Layers
14:54 - Training Data and Detailed Scene Descriptions
21:45 - Synthetic Data and Diversity in Design Generation
22:53 - Evaluating Image Generation Models and Design Editing
25:00 - Vision-Language Models for Evaluating Design Quality
26:30 - Tuning Model Taste and Visual Style
31:13 - Canva Brush
35:49 - Simplifying Agent Harness
37:36 - Code Generation
39:44 - Optimizing Inference Stack
47:54 - Distilling Craft for Speed and Cost
50:24 - Canva Models Available Through MCP and APIs
54:23 - What’s Next for Canva Research
Learn from AI experts, share challenges and bottlenecks in your projects, exchange ideas with fellow AI enthusiasts, and more in our Generative AI study group! All skill levels are welcome to join. Just register at twimlai.com/community/ and we’ll see you in an hour!
1
1
368
AI can help scientists write code, analyze data, and draft papers. But can it generate the ideas that lead to new discoveries? @RichardSocher, former chief scientist at Salesforce, founder of @youdotcom, and co-founder of @Recursive_SI, joins us to explore the move from AI assisting researchers to AI conducting research. Starting with AI improving AI, we examine what it takes to propose ideas, run experiments, and validate results, and how those capabilities could eventually accelerate discovery across biology, chemistry, and physics.
The conversation covers Recursive’s reported results in model training and GPU kernel optimization, why fast verification matters, and how reward hacking complicates the search for real improvements. We also discuss the data and experimental infrastructure needed for broader scientific discovery, whether commercial incentives can keep increasingly capable systems aligned, and Richard’s vision for a “Eureka machine.”
🗒️ Full show notes: twimlai.com/go/783.
📖 CHAPTERS
===============================
00:00 - Intro - Richard Socher and the Eureka Machine
03:11 - AI for Science Examples across Physics, Chemistry, and Biology
05:47 - Organoids, Virtual Cells, and Drug Repurposing Case Studies
10:16 - Evaluating AI for Science Timelines, Hype, and Physical Bottlenecks
13:47 - Four Pillars of the Eureka Machine and AI Scientist Agents
16:37 - Open-Endedness and Environments for Scientific Ideation
21:04 - Guided Search, Reward Engineering, and Reward Hacking
26:02 - Can AI Truly Innovate? Recursive Self-Improvement in Practice
29:25 - Recursive’s Reported Results on NanoChat, NanoGPT, and CUDA Kernels
33:02 - Verification Speed, Long-Horizon Attribution, and System Generalizability
37:49 - Scaling Compute, Biological Data, and Mitigating Reward Hacks
40:11 - Commercial Incentives and AI Alignment
41:46 - AI Pacing Debates and Regulating Applications vs. Intelligence
44:29 - Dual-Use Science Risks, Accountability, and the Case for Progress
Just one hour away before our Generative AI study group starts! Exchange effective generative AI strategies, collaborate on AI projects, share news on the latest AI developments, and more! We’ll see you in a bit!
Join our Generative AI study group and explore topics like long-horizon agentic coding, spatial computing, Mixture-of-Experts (MoE) routing techniques, and more! Simply register at twimlai.com/community/ and join tomorrow’s session at 8 am PT.
342
Join our community of AI enthusiasts in our Generative AI study group every Friday at 8 am PT. Dig into token-level knowledge distillation, explore latent-space world models, discuss continual learning techniques, and more! Register at twimlai.com/community/ to join the conversation.
242
AI agents are becoming accessible to everyone, raising the bar for developers building the next generation of agent products. When capable, general-purpose agents are available out of the box, differentiation increasingly comes down to the architectural decisions and specialized experiences that make an agent uniquely valuable.
In this episode, @Dan_Jeffries1, creator of @nautiloapp, joins us to discuss what he's learned building a platform for personal and collaborative AI agents. We explore his approach to persistent memory, coordinating work across people and agents, managing private context, and designing deterministic tools that complement LLM capabilities. Dan also shares his views on multi-agent architectures, computer use, and the tradeoffs between local and cloud models.
The conversation offers a practitioner's perspective on where the opportunities lie as AI agents become more capable, more accessible, and more difficult to differentiate.
🗒️ Full show notes: twimlai.com/go/782.
📖 CHAPTERS
===============================
00:00 - Introduction
02:01 - The agency problem
07:37 - Nautilo
13:46 - Multi-agent systems
19:42 - Context window limitations
22:21 - Tool design philosophy: "Magical Hammer Design”
27:09 - Venice
28:48 - Local models and memory constraints
35:33 - Balancing privacy with model selection
38:24 - Bring your own keys and subsidized compute
40:39 - Co-creative tools
45:40 - Building agent products against “the wall”
49:08 - Bespoke software and the challenge of defensible moats
We’re just an hour away from the start of this week’s session of our Generative AI study group! Dig into agentic engineering, discuss KV-cache compression techniques, explore latent Mixture of Experts (MoE), and more! Make sure you’ve joined our Slack community after signing up on twimlai.com/community/. We’ll meet you there!
1
402
🕗 Tomorrow at 8 am PT, our Generative AI study group meets to share insights on the latest AI developments, discuss sparse coding techniques, trade model evaluation strategies, and more! No experience required—just sign up at twimlai.com/community/, and we’ll see you tomorrow at 8 am PT.
330
Our weekly Generative AI meetup occurs every Friday at 8 am PT. Register at twimlai.com/community/ and dig into Heavily Compressed Attention (HCA), exchange prompt engineering tricks, explore video-language reasoning models, and more! The discussions are informal and participants are free to join or leave at any time during the session.
AI agents need the institutional knowledge experienced employees rely on: why a system works a certain way, which policies apply, and what never made it into the documentation. Capturing that knowledge is hard. Keeping it useful as people and agents add more may be harder.
In this episode, @Ahmad_Al_Dahle, CTO of @Airbnb, shares how his team is building Project Everest, a “company brain” that captures knowledge from employees’ workflows for AI agents. We explore stale information and context poisoning, why too much context can backfire, and how the team evaluates whether shared knowledge improves engineering work.
We also discuss measuring AI productivity beyond token usage and PR counts, the limits of chat-based interfaces, and deploying support agents in a marketplace where resolving disputes means mediating between conflicting interests.
🗒️ Full show notes: twimlai.com/go/781.
📖 CHAPTERS
===============================
00:00 - Intro
01:35 - Three Pillars of AI at Airbnb
03:19 - AI for Engineering Productivity: Balancing Velocity and Craft
06:20 - Project Everest
10:30 - Everest architecture options: databases, files, and retrieval
12:25 - Stale context, context poisoning, and progressive disclosure
13:14 - AirChat and a multi-model coding stack
14:38 - Engineering evals and calibrating model judges
17:37 - Managing model costs without token quotas
20:12 - Measuring engineering value beyond tokens and PR counts
24:12 - Behind-the-scenes agents for marketplace trust and quality
26:36 - Adversarial support mediation: evidence, policies, and conflicting interests
30:33 - Deployed support agents and limits requiring human judgment
33:05 - Support workflows and choosing human escalation
36:05 - Natural language search without a chat-first interface
39:41 - External agent access, MCP, and travel trust
44:09 - Building agent frameworks versus adopting open source
45:55 - Post-training small models for specialized, low-latency tasks
48:25 - Scaling Everest and unresolved agent experience challenges
50:54 - Conclusion
1
467


