AI and Platform Engineering (Category AI)
An interesting talk by Igor Maslov, VP of Coretech & Data at T-Bank, opening the Platform Engineering Night conference I mentioned earlier. Igor discussed how AI affects engineers’ work and the development of engineering platforms. Here are the main points.
1. AI strengthens existing platforms The immediate task is to add AI to existing engineering workflows, from creating pipelines to analyzing telemetry. Google’s Measuring Developer Goals whitepaper is useful for seeing which scenarios it identifies; I covered it before. There is also episode nine of Research Insights Made Simple, where my colleague Kolya Bushkov and I discussed Google’s What Do Developers Want From AI?
2. Cognitive load and lost skills Igor notes that intensive AI use leads to the loss of low-level skills, but sees this as normal evolution. Few people write assembly now, and most engineers no longer understand it, without much concern. Similarly, we may write Java with assistants, while some people lose the ability to write or read it unaided :)
3. Copilot mode as the baseline People remain responsible for the final result, minimizing the risks of model hallucinations. Anthropic applies this philosophy in Claude, where the model assists rather than replaces the developer. This matters particularly in critical systems such as finance and healthcare, where AI errors are unacceptable.
4. The future: AI as platform engineering In the long term, AI could become the main interface for engineering processes, removing the need for traditional tools. This resembles Kubernetes becoming the container-orchestration standard, although AI has not yet had a comparable moment.
5. Vibe coding and personalized software Generating code for specific business cases or individual user needs is a major trend. Igor warns that these systems require careful validation to avoid maintainability problems.
6. Freeing developers from routine work He suggests that 80% of experimental work can be automated, freeing time for more important tasks. OpenAI’s Code Interpreter can iteratively solve complex programming and data-analysis problems.
7. ML platforms: flexibility and readiness for change Investment in ML platforms makes sense, but requires readiness for technological shifts. For example, the transition from traditional neural networks to transformers in the 2020s forced many companies to rethink their infrastructure completely. Igor emphasizes keeping platforms modular so they can adapt to new algorithms and hardware.
8. AI’s missing Kubernetes moment There is no universal AI-platform standard yet, and he suggests waiting for it to emerge. Companies have different approaches: OpenAI with Agents SDK, Anthropic with a security focus, and Google with comprehensive AI solutions.
9. The high-risk nature of current AI R&D Modern AI solutions demand substantial resources and mainly suit large companies, as illustrated by OpenAI, Google, and Microsoft’s large platform investments.
10. Gradual adoption The recommendation is to improve tools first and hold off on complex low-code processes. Anthropic follows a similar philosophy, providing powerful models with an emphasis on controlled adoption.
Igor gave an excellent talk. His central message is that an AI revolution in platform engineering is inevitable, but calls for a measured approach focused on gradually improving existing processes rather than replacing them wholesale.
#Management #Leadership #Software #SoftwareDevelopment #Metrics #Devops #Processes #AI #ML #DevEx