A DotNet developer’s note on Technology Radar V28

As a new Technology Radar release with a bunch of interesting techs, a few drew my attention. Techniques
  • Applying product management to internal platforms: product mindset includes a roadmap, value to biz, and consumer experience enhancement.
  • CI/CD infrastructure as a service: Our team has been using Jenkins hosted on-prem and started moving to Azure DevOps. It seems like a no-brainer. However, as the CI/CD runner can access source code and credentials, zero trust security must be implemented.
  • Dependency pruning: Just a good practice to follow in this risker world.
  • Run cost as architecture fitness function: Estimating cost change in advance of architectural change will be the way to avoid unexpected cost surges.
  • Bounded low-code platforms: Use low-code tools for simple tasks only. After having been using Nintex K2 no-code tool for over one year now, this is exactly what I felt. Other low-code solution examples are Mendix or Microsoft Power App.
  • Demo frontends for API-only products: This may help to get stakeholders to head around the biz values of API products.
  • Lakehouse: Data Lake and Data Warehouse in one solution. e.g. Databricks.
  • Verifiable Credential:
  • API-aided test-first development: Ask ChatGPT with the prompt of architecture and tech spec to generate an implementation plan for a feature. Then ask to test for it with a given acceptance criteria. The developer can implement a working code from there.
  • Domain-specific LLMs: e.g. OpenNyAI for legal
  • Prompt Engineering: creating a effective prompt for generative AI to get a best result.
  • Reachability analysis when testing infrastructure: e.g. Azure Network Watcher
  • Self-hosted LLMs: e.g. llama.cpp, GPT-J, GPT-JT, LLaMA
  • Tracking health over debt: Rather than blindly tackling technical debt, assess technical health like development, operations and architecture. When the health becomes red, do some action. e.g. What good software looks like at REA | REA Group Ltd (rea-group.com)
  • Zero trust security for CI/CD: CI/CD agents will have powerful access across the company system. It is worth giving extra security care. e.g. minimal access, access through OIDC, etc
  • Webhooks need to be treated as credential.
  • Lambda pinball architecture: Serverless architecture can be challenging to debug. Use it for the right place.
  • Make some buffer when planning. Otherwise you will see what's happening on congested highways.
Platforms
  • Contentful: Headless CMS. API-first approach. Separate the backend from the presentation.
  • GitHub Action: Not sure which one to pick between GitHub Action and Azure DevOps.
  • Dapr: Distributed Application Runtime. Alternative to Dotnet Akka.
  • Immuta: data security platform.
  • Matter: open standard for smart home tech.
  • Passkeys: login without password
Tools
  • FOSSA: Open Source License Compliance
  • Gitleaks: open-source static application security testing, which detects hardcoded secrets.
  • Mend SCA: finds vulnerable open-source dependencies.
  • Mozilla SOPS: encrypt secrets in text files like app.config.
  • Soda Core: open-source data quality and observability tool
  • Steampipe: an open-source tool to query clouds with SQL.
  • TruffleHog: open-source static application security testing
  • Typesense: open-source, typo-tolerant search engine. High performance with small index size.
  • Vite: a front-end build tool. Fast. Getting more traction.
  • ChatGPT
  • GitHub Copilot
Languages & Frameworks
  • PyTorch: go to ML framework over TensorFlow.
  • Mikro ORM: Typescript centric ORM.
  • Stencil: library to build reusable components that work for any framework.
  • Synthetic Data Vault: generate data for testing environment. The SDV uses a variety of machine learning algorithms to learn patterns from your real data and emulate them in synthetic data.
  • .Net7 Native AOT: deploy with dependency on IL or JIT. This makes cold start really quick.
  • .Net MAUI: Good candidate to upgrade from Xamarin. But not mature enough yet.
  • nanoGPT: a framework for training medium-size generative pretrained transformers(GPT). Good to experience the building blocks of the GPT architecture.
  • Qwik: JS framework for quick first-time website loading.

Leave a comment

Please note that we won't show your email to others, or use it for sending unwanted emails. We will only use it to render your Gravatar image and to validate you as a real person.