High-throughput AdTech
Low-latency bidding, OpenRTB pipelines, budget pacing, and fraud checks at scale.
Introduction Collection
GC algorithms explained with JVM use-cases and tuning tips.
Building reliable systems,
sharing what I learn
Technology leader with 10+ years of experience. Focus on JVM internals, AdTech infrastructure, distributed systems, and data pipelines. I enjoy turning complex ideas into practical guides.
- JVM · GC
- AdTech · RTB
- Distributed Systems
- Kafka
- Cassandra
Table of Contents
What I work on
Distributed storage
Data modeling and consistency trade-offs with Cassandra, Postgres, and object stores.
JVM internals
Tuning garbage collectors, memory profiling, safe concurrency, and performance budgets.
Engineering leadership
Growing cross-functional teams, roadmaps, and pragmatic delivery processes.
Featured articles
JVM
AdTech
How Bidding Works
From SSP to DSP and OpenRTB — latency budgets that actually matter.
Data
Introduction to Cassandra
Modeling for queries, not tables. Reads, writes, and consistency trade-offs.
AdTech
Learn AdTech
A practical landscape of players, protocols, and pitfalls for newcomers.
Reviews
Artur Mkrtchyan consistently brings clarity to complexity. In a
world drowning in buzzwords and shallow claims, his insights into
software testing, process design, and quality culture stand out
for their depth and practicality.
— Elena Ivanova, Head of QA at TechSphere
I’ve collaborated with many thought-leaders in the testing domain,
but Artur’s approach is different: he doesn’t just preach theory —
he offers paths I can adopt tomorrow, in my team, my product.
— Michael Ong, Quality Architect & Consultant
Whether it’s risk-based testing, test
automation strategy, or crafting a sustainable quality mindset, Artur’s
recommendations feel grounded, field-tested, and free of fluff.
That’s rare.
— Sara Kim, Director of Quality Engineering
Artur’s influence in the QA community in Armenia and beyond is
visible: his work in training, organization founding, and
real-world consulting shows he’s not just an idea person — he
delivers.
— Anaïs Moreau, Software Testing Evangelist
They don’t just teach you what works today — they build toward
what will last. Their approach to state, forms, and testability
sets a standard.
— Ravi Chandra, Software Architect
Highlights
-
2025
Led migration to Kafka-centric event backbone; reduced p99 latency by 37%.
-
2024
Built RTB bidder handling 150k RPS with adaptive pacing & budget guards.
-
2023
Introduced GC tuning playbook; cut allocation rate hotspots by 45%.
Projects & guides
GC Visualizer
Small tool to compare GC pauses across workloads; export heatmaps.
RTB Sandbox
Local dev environment to test OpenRTB flows with synthetic traffic.
Data Modeling
My patterns & anti-patterns for read-optimized schemas.
Tracing Toolkit
A lightweight toolkit for distributed tracing with ready-made dashboards and instrumentation snippets.
Engineering Principles
Latency Budget First
Every feature is designed against strict p95/p99 latency goals with alerts and graceful degradation.
Design for Failure
Idempotency, retries, circuit breakers, and backpressure are built into critical paths.
Observe Everything
Metrics, tracing, logging facades, and SLO/SLI monitoring with real traffic samples.
Simplify Aggressively
Minimal moving parts, explicit contracts, and avoiding unnecessary abstractions.
Data Over Guesses
Feature flags, experimentation, and A/B comparisons based on real-world costs.
Docs as Interface
Concise “how-to” notes near the code, flow diagrams, and SLA on a single screen.
Case Studies
Bidder Latency Cut
- −37% p99
- +22% win-rate
- 150k RPS
Migrated critical path to batch-fetch + zero-copy serialization. Isolated cold branches from hot path.
Kafka Backbone
- 3× throughput
- 0 DLQ steady
- 7 domains onboarded
Unified event backbone with schema contracts, retention policies, and load isolation per topic.
GC Tuning Playbook
- −45% allocation
- −28% pauses
- ZGC rollout
Profiling allocations, pooling where appropriate, and migration to ZGC for sub-10ms pauses.
Streaming Analytics
- +4× query speed
- −35% infra cost
- 99.99% uptime
Rebuilt the real-time analytics pipeline with ClickHouse + Kafka, optimizing rollups and reducing infrastructure footprint while maintaining strict SLA guarantees.
Speaking & Mentoring
Conference Talks
Delivered talks on JVM internals, real-time bidding pipelines, and distributed data modeling at international conferences.
Workshops
Hands-on workshops covering GC tuning, Kafka event design, and scaling adtech platforms to 100k+ RPS.
Mentoring
Helping engineers grow in system design, performance debugging, and career direction through 1:1 mentoring.
Open Source
Contributing to open-source projects and sharing internal tools with the community to encourage knowledge exchange and collaboration.
Frequently Asked Questions
I focus on JVM internals, garbage collection, AdTech infrastructure, distributed systems, and data engineering. Most of my writing comes from real-world lessons scaling high-throughput platforms.
Yes — I enjoy working with engineering teams on system design, performance debugging, and architecture reviews. If you have an interesting challenge, feel free to reach out.
Absolutely. I mentor engineers on topics like distributed system design, JVM tuning, and career growth. I’ve also run workshops on GC tuning, Kafka event modeling, and scaling AdTech pipelines.
I’m based in Berlin, Germany, and usually work with international teams across Europe and the US. Most of my collaboration is fully remote.
Yes — I’ve spoken at several industry conferences on JVM internals, real-time bidding, and distributed data modeling. I also share materials from my talks here on the site.
The easiest way is via email at mkrtchyan.artur@gmail.com. You can also connect with me on LinkedIn or GitHub — links are in the footer.