November 08, 2021

A New Quantum Computing Method Is 2,500 Percent More Efficient

Today, most quantum computers can only handle the simplest and shortest algorithms, since they're so wildly error-prone. And in recent algorithmic benchmarking experiments executed by the U.S. Quantum Economic Development Consortium, the errors observed in hardware systems during tests were so serious that the computers gave outputs statistically indiscernible from random chance. That's not something you want from your computer. But by employing specialized software to alter the building blocks of quantum algorithms, which are called "quantum logic gates," the company Q-CTRL discovered a way to reduce the computational errors by an unprecedented level, according to the release. The new results were obtained via several IBM quantum computers, and they also showed that the new quantum logic gates were more than 400 times more efficient in stopping computational errors than any methods seen before. It's difficult to overstate how much this simplifies the procedure for users to experience vastly improved performance on quantum devices.

Design Patterns for Machine Learning Pipelines

Design patterns for ML pipelines have evolved several times in the past decade. These changes are usually driven by imbalances between memory and CPU performance. They are also distinct from traditional data processing pipelines (something like map reduce) as they need to support the execution of long-running, stateful tasks associated with deep learning. As growth in dataset sizes outpace memory availability, we have seen more ETL pipelines designed with distributed training and distributed storage as first-class principles. Not only can these pipelines train models in a parallel fashion using multiple accelerators, but they can also replace traditional distributed file systems with cloud object stores. Along with our partners from the AI Infrastructure Alliance, we at Activeloop are actively building tools to help researchers train arbitrarily large models over arbitrarily large datasets, like the open-source dataset format for AI, for instance. ... Even though the problem of transfer speed remained, this design pattern is widely considered as the most feasible technique for working with petascale datasets.

Daily Standup Meetings are useless

Standup are not about technical details, even though some technical context can help to frame the arisen complexity and allow PjM and Leads to take the necessary steps to enable you achiving your tasks ( additional meetings, extending the deadline, reestimating the task within the sprint, set up pair programming sessions etc). According to the official docs the Daily Scrum is a 15-minute event for the Developers of the Scrum Team. The purpose of the Daily Scrum is to inspect progress toward the Sprint Goal and produce an actionable plan for the next day of work. This creates focus and improves self-management. Daily Scrums improve communications, identify impediments, promote quick decision-making, and consequently eliminate the need for other meetings. Honestly I am not so sure about the last point, due to the short time allocated to it, it can indeed generate other meetings, follow-ups between Tech Lead and (some of ) the developers, or between the team and the stakeholders, or among developers which decide to tackle an issue with pair programming. 

When Is the Waterfall Methodology Better Than Agile?

Is waterfall ever better? The short answer is yes. Waterfall is more efficient, more streamlined, and faster when it comes to specific types of projects like these: Generally speaking, the smaller the project, the better suited it is to waterfall development. If you’re only working with a few hundred lines of code or if the scope of the project is limited, there’s no reason to take the continuous phased approach; Low priority projects – those with minimal impact – don’t need much outside attention or group coordination. They can easily be planned and knocked out with a waterfall methodology; One of the best advantages of agile development is that your clients get to be an active part of the development process. But if you don’t have any clients, that advantage disappears. If you’re working internally, there are fewer voices and opinions to worry about – which means waterfall might be a better fit; Similarly, if the project has few stakeholders, waterfall can work better than agile. If you’re working with a council of managers or an entire team of decision makers, agile is almost a prerequisite. 

To Secure DevOps, Security Teams Must be Agile

Focusing on a pipeline using infrastructure-as-code allows security teams to build in static analysis tools to catch vulnerabilities early, dynamic analysis tools to catch issues in staging and production, and policy enforcement tools to continuously validate that the infrastructure is compliant, Leitersdorf said. "If you think about how security can be done now, instead of doing security at the tail end of the process ... you can now do security from the beginning through every step in the process all the way to the end. Most security issues will be caught very early on, and then a handful of them will be caught in the live environment and then remediated very quickly," he said. Developers get to retain their speed of development and deployment of applications and, at the same time, reduce the time to remediate security issues. And security teams get to collaborate more closely with DevOps teams, he said. "From a security team perspective, you feel better, you feel more confident, you have guardrails around your developers to reduce the chance of making mistakes along the way and building insecure infrastructure and you now have visibility into their DevOps process, a huge bonus," Leitersdorf said.

How to Avoid Vulnerabilities in Your Code

In general, the "minor problems" are precisely those that result in the most extensive security disasters, and we can say that failures usually present two gaps: We encountered flaws such as insecure code design, injection, and configuration issues through a code vulnerability intentionally or unintentionally: within the TOP 10 demonstrated by OWASP; Through the vulnerability of operations: among the most common problems, we can mention the choice for "weak passwords", default or even the lack of password. A second common failure is the mismanagement of people's permission to a document or system. These types of problems, unfortunately, are pretty standard. Not by chance, 75% of Redis servers have issues of this type. In an analogy, we can say that security flaws are like the case of the Titanic. Considered one of the biggest wrecks, most people are unaware that the ship had a "small problem": the lack of a simple key that could have opened the compartment with binoculars and other devices to help the crew visualize the iceberg in time and prevent a collision.

Read more here ...