From Zero to Zero-Code Observability: A Journey with OpenTelemetry

Abstract

Our customer, a fast-growing neuroscience startup, was scaling rapidly across multiple regions and environments. Observability wasn’t a priority at first. Their architecture had already evolved from monoliths to more than 15 microservices, but debugging still relied heavily on CloudWatch logs and manual guesswork.
We had just completed their EKS migration and rolled out CI/CD with GitOps. The next big challenge was clear: introduce observability without slowing down teams that were already stretched. Developers weren’t keen on making code changes. Leadership wanted to see value quickly. We had to prove that observability could be seamless, low-maintenance, and worth it.
With OpenTelemetry and a zero-code approach, we delivered full visibility across systems. No code rewrites. Just smart instrumentation and the right integrations. This talk is about how we made that happen.

The Burning Problem

1. Developers didn’t want to touch application code to enable observability
2. The stack was a mix of Java, Scala, and Python, with no consistent patterns
3. ALB metrics from CloudWatch couldn’t be pulled directly into OpenTelemetry pipelines.
4. Leadership needed a working demo before moving forward

What We Delivered

1. A unified observability layer using OpenTelemetry and Signoz
2. Logs, metrics, and traces correlated across environments with minimal effort
3. Infrastructure telemetry from services like AWS ELB, using adapted CloudWatch metrics from CloudWatch Exporter for Prometheus.
4. A modular setup that could support any OpenTelemetry-compatible backend.
5. All of this with little to no changes to service code

My Role in the Project

This was my first real-world project as a Site Reliability Engineer, and it came with high expectations, not from others, but myself. I was responsible for figuring out how to get the core RED metrics with as little effort as possible. I planned the early rollout in lower environments and worked on connecting logs, metrics, and traces in a way that made sense across different systems. It was hands-on, fast-moving, and full of learning moments.

Takeaways from the Talk

1. How to use OpenTelemetry agents for zero-code instrumentation across different languages
2. What it takes to unify observability in fragmented microservices without slowing teams down
3. How to navigate mindset shifts when introducing observability in high-pressure environments?
4. What it’s like to drive impact as an early-career SRE on a real production problem

Audience

1. SREs, platform engineers, and DevOps teams trying to roll out observability in fast-moving setups.
2. Engineering leaders looking to add visibility without adding friction

Bio

I’m an SRE working at the intersection of infrastructure and developer experience at One2N. I focus on observability, automation, and building dev tools. I’m pretty new to the landscape, and it never fails to amaze me. Previously, I had worked on building integrations for Postgres Clusters for Runwhen.com. Apart from being a script kiddie in tech and a cyber janitor, I love to hack around with Linux and headbanging to Metal music (sometimes both simultaneously).