Modern software systems release “machine data” (logs, metrics, etc.) that are critical to detecting and understanding the abuse, but the size and complexity of this data far exceeds the human ability to perform the necessary analysis and capture.
Timely action. For this reason, we see a lot of possibilities to create automated systems that analyze (and work on) these machine learning data to improve the security, performance, and reliability of financially complex software services.
There is also a lot of exciting research around “ML on code”: automated detection of risky pull requests, automatic bug localization, intelligent IDE help, and so on. Given the well-known challenges of building and operating software systems, there is much room for improvement over the entire lifecycle. Overall, I think we are going through a really interesting time to apply ML techniques to software development, security, and operations.
ML has tremendous opportunities in continuous automation of bug fixing, testing, deployment, and code optimization.
There is no possibility of replacing the human factor in software development. ML could not decide which was right or wrong. It will continue to identify more tests that can be automated. By solving and automating small tasks daily to make intelligent decisions, you will be able to deliver more, faster, more quality, and with less human involvement.
The great promise lies in the pace of development and production — enabling us to do so much more with our time. The simplicity with which we introduce feedback and iterative cycles has allowed us to redirect and focus on outcome-led programming. Capacity beyond what humans can achieve in software development.
There are also many possibilities for making the process of creating and producing software very fast. But for me, the opportunity lies in the possibility of humans and machines working together — moving the role of the programmer, developing new skills and freeing them to focus on what they are good at, and allowing the machines to be mundane.
There are a lot of areas in tech that are going to see huge improvements from ML in the coming years, but I am very excited about Discoverability — the process of finding a product or experience. From grocery shopping to finding flights, searching for information on Google, we all spend many hours a week doing these things.
Teaching our priorities to the computer can help us do these things faster. We get more free time, and in contrast to sectors like self-driving cars, no one will lose a job. It is a pure universal good, and a machine learning helps.
Next-generation applications that use ML are seamlessly integrated and are in the fabric of the ML app, so ML is working on real-time data, re-training, and testing, and decision-making. Develop an integrated platform that integrates ML into the data platform to power the required data size.’
2. Using Artificial Intelligence to detect COVID-19
3. Real vs Fake Tweet Detection using a BERT Transformer Model in few lines of code
4. Machine Learning System Design
Allowing humans to focus on things like creativity and non-linear thinking requires problems. Automate repetitive tasks. AI services are what builds intelligence. Use AI to automate specific tasks around cleaning and preparing data and dashboard creation. How to deal with technology without BI training.
More data scientists. With more open source ML libraries it is more accessible to software developers.
There seems to be a great deal of potential for combining classical robotic algorithms with ML. ML improves the performance of some of the algorithms while maintaining the transparency of the original method.
DevOps can be a big winner. This becomes part of the software development process as people begin to find value in the Ops data. Developers use tools if they are part of the toolchain everyone uses. If ML-based data is part of the feedback loop of the dev pipeline and code processes they are likely to use it. Subtle improvement in quality that touches the intended features. They can see change for the better or worse over time.
ML SaaS-if is even more so. Some common problems come with the best models and people are licensed and tailored to their use case. We stop building ML models from scratch. We bring the best race to this is for application.
I believe that outside of most profitability, there is a vertical-centric approach. It is difficult to build a one-size-fits-all horizontal solution, and successful companies can spend their time in a column and make money from a specific problem.