A.I Engineering Sign up at https://xitter.cf/t.co/NSrYj7OYKi
London, England
Joined September 2019
- Tweets3.1K
- Following137
- Followers151
- Likes4.3K
ML Ops retweeted
We Must Pace the Frontier: I’ve written a new essay on why the AI industry should slow down, with a three-part plan for doing so.
Anthropic is unilaterally committing to the first of these steps. We’ll provide third-party evaluators with permanent, employee-level access to our systems, so that they can verify adherence to our safety measures, report on incidents, and assess models’ alignment during training.
You can read the full post here: darioamodei.com/post/we-must…
ML Ops retweeted
We've been awarded £708m in a contract extension to progress the concepting and development of next generation combat air technologies, underpinning the UK's role on the Global Combat Air Programme and allowing us to develop capability beyond the sixth-generation aircraft.
Read more 👉baes.co/HN6C50ZrfMx
ML Ops retweeted
📽️The people behind the Demonstrator explain what it means, and why it matters.
This is the country's first crewed combat air flying demonstrator in four decades.
A once-in-a-generation national endeavour, and a real test of what British industry can deliver, together.
Proving advanced technologies before they reach the front line, at the same time as developing the national skills that Britain needs for the future of combat air.
Watch the full video now 👇
baes.co/3rVZ50ZyWRU
ML Ops retweeted
In F1, the data can't be late.
Michael Taylor explains how the Mercedes-AMG PETRONAS F1 Team streams car data to Azure and creates virtual sensors for split-second engineering decisions: msft.it/6012aViZg
ML Ops retweeted
You know you are hitting the API and making round trips again and again when a user asks the same question, like "what's the refund policy?" or "what are the conditions of a refund?"
You might not have a cache layer built in, or if you do it's probably keyword-based, which only works on exact matches. In that case semantic cache can be useful.
What semantic cache does is store previously asked questions along with their answers. When a new query comes in, it checks for semantic similarity against cached questions. If the similarity score is above a threshold, it returns the cached answer without going through retrieval and generation again. Same answer, no unnecessary API calls or retrieval round trips.
You can learn more about how semantic cache works in practice in our guide: qdrant.tech/articles/semanti…
ML Ops retweeted
Cost per token dropped ~80%, then reversed. Josh Collier, FinOps Lead at @Superhuman, on why production LLM bills are trending up - model prices doubling and a ~10% US data residency premium teams miss.
ML Ops retweeted
We must build the academia-industrial pipeline needed to sustain Africa's drone industrial base
With Nigeria's Parliament recent motion on the need for a National Drone Industrial Policy, Terra-Miva partnership will fill the talent gap
Terra Industries is partnering with @mivauniversity to build integrated robotics and drone labs across Miva learning centers, starting with a pilot in Abuja.
Students will learn to design, assemble, test, and fly drones as part of their academic programs. Terra will run hands-on workshops, train faculty, and offer internships and jobs to top performers.
Africa's next generation of engineers should learn by building. This is how we develop the talent that will build the continent's industrial base.
ML Ops retweeted
What happens when the access gap actually closes? This is what it produces.
#AccessGapClosed #Learn2Earn #FutureAfrica #InnovationInEducation #EmpowermentThroughKnowledge
ML Ops retweeted
What happens when African talent gets access? Here's our answer.
#AfricanTalent #AccessToOpportunities #Empowerment #Innovation #FutureLeaders
ML Ops retweeted
For forty years, transactional and analytical databases were treated as separate systems, connected by pipelines that moved data between them. As AI agents generate more database activity, that architecture is being reconsidered around speed, scale, and a single copy of governed data.
Databricks Co-Founder and Chief Architect @rxin joined @DataCamp’s DataFramed podcast to discuss LTAP, Lakebase, real-time analytics, governance for AI agents, self-service analytics with Genie, and more.
Watch the full conversation: youtube.com/watch?v=US6Z_buK…
ML Ops retweeted
Before, running a remote MCP server meant managing session state, which limited where you could run it.
Now that MCP is stateless, you can deploy on serverless and edge infrastructure, or scale horizontally behind any load balancer.
ML Ops retweeted
Digital education needs internet and devices.
Better skills need real employer demand.
Global employment needs payments and policy that let people work across borders.
Talent can be everywhere. The conditions for it to thrive have to be built.
ML Ops retweeted
On-device AI has three advantages over cloud. Cost, privacy, and latency.
Alan Zhu from @Qualcomm demonstrated why latency is the most underestimated advantage of on-device AI at Vector Space Day SF.
Alan ran the same agentic search task on NPU, GPU, and CPU. NPU responded instantly. GPU took 12 seconds. CPU took 30, then throttled down as the device heated up.
The NPU held 90 tokens/sec for the full session at 37°C. The CPU started lower and kept dropping.
The reason: NPU prefill speed. Fast enough to process thousands of local files and return a cited answer in seconds, entirely on device, no network required.
That's what makes local AI agents actually practical, not just theoretically possible.
Full talk: youtube.com/watch?v=FlAmmVSY…
ML Ops retweeted
TLMs: Tiny LLMs and Agents on Edge Devices with @cormacb
youtube.com/watch?v=-TiET_K-…
Function Gemma ships at 270 million parameters and runs nearly 2,000 tokens per second prefill on a Pixel 7. Out of the box, it hits 46% accuracy on a fixed set of app intents. Fine tune on a synthetically generated dataset and that clears 90% on eight of ten functions.
Cormac walks through the two paths developers have for on device AI: a skill harness built on Gemma 4 with a restaurant roulette demo running fully on device. Then Eloquent, a production transcription app built by chaining two sub billion parameter models together.
cc @osanseviero
ML Ops retweeted
For my first post, I’m sharing a letter @NVIDIA signed on why open models matter.
AI will transform every industry, power every company, and be built by every country.
Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty.
The world needs both frontier closed models and frontier open models.
images.nvidia.com/pdf/Open-W…
