I wanted a concrete number for whether AI tools were actually helping my work. Aggregate productivity metrics couldn't give me one. Here's the approach that finally did β and what it showed me.
Measuring AI Adoption Through Work Artifacts
Hey! π I'm Dzmitry Martavoi, Co-CTO at IDT.
I architect and develop intricate distributed systems with a focus on backend development, identity & access management (IAM), and AWS Cloud infrastructure. Passionate about event-driven architectures, microservices, OAuth/OIDC, and building high-performance financial systems.
Tech Stack:
βββ Languages: Go, C#/.NET, C++
βββ Architecture: Microservices, Event-Driven, DDD
βββ Auth/IAM: OAuth2/OIDC, Ory Kratos, Ory Hydra, AWS Cognito
βββ Cloud: AWS
βββ Data: Kafka, MongoDB, Couchbase, Snowflake
βββ DevOps: Containers, IaC (Terraform/AWS CDK), GitHub Actions
βββ APM: Grafana, Prometheus, OpenTelemetry, NewRelic
βββ AI: MCP, Cursor AI, Claude Code
π Check out my resume for full details
π¬ Reach me: Telegram |
LinkedIn |
GitHub
π€ View my Modern C++ Presentation
I wanted a concrete number for whether AI tools were actually helping my work. Aggregate productivity metrics couldn't give me one. Here's the approach that finally did β and what it showed me.
How to build production-grade, fine-grained authorization in 2026 using Ory Keto β the OSS implementation of Google's Zanzibar model.
How to add OAuth2.1 authorization code + PKCE, OIDC id/access/refresh tokens, and modern security standards to an existing user base β without replacing it β using Ory Hydra.
Why building custom auth is a mistake in 2026βand how Ory Kratos delivers flexible, schema-driven identity with pre/post hooks, phone/SMS, OIDC, MFA, and more
A developer-friendly guide to storage types: why row storage (OLTP) and columnar storage (OLAP) exist, and when HTAP makes sense