Aditya Reddy

Frontend · Platform · Kubernetes

Senior software engineer. I build and run production systems for companies where downtime has a dollar sign attached to it.

Started in the pixels.

Ten years of shipping interfaces people actually touch. The label that stuck at work is frontend guru, though these days the pixels are only half the job.

Then the stack underneath.

Backend, cloud, pipelines, security. Each layer cause the one above it kept breaking on something I couldn't see yet.

Averages hide the users who are waiting.

The distribution has a tail, and the tail is where the experience actually happens. p95 says someone is hurting. p99 says who.

One keystroke. 300 re‑renders.

A form-heavy enterprise app cloned its whole state tree on every keystroke. Every point of light is a component doing pointless work.

Rewrite the state layer and it's one.

Instrumented numbers, not an estimate.

INP 426ms → 268ms

Now the platform under all of it.

Kubernetes, Vault, Terraform, the release trains. I run what I build, and I carry the pager for it.

Faster in the field.

A controlled experiment proved which change did it. Making it fast is half the job. Proving why is the half people skip.

p95p99
About

The seam between the code and the platform.

Aditya Reddy on the Brooklyn Bridge
Brooklyn Bridge, New York

Right now I lead observability for a major application rebuild at Northwestern Mutual. The legacy and rebuilt stacks run side by side, and the monitoring has to referee between them fairly or the comparison is worthless. I defined the SLI framework the stakeholders signed off on, and moved measurement to the gateway instead of trusting what each application reports about itself.

Before that I was a core engineer on straight-through processing of insurance applications: 16 Node microservices, 3 React apps, an event-driven architecture on AWS. I took the team's DevSecOps role cause someone had to, and it turned out I liked the pipelines as much as the product. PagerDuty on-call too. You should carry the pager for what you build though.

Nine years in software, seven of them building and operating production web systems in the US, across financial services, healthcare and retail supply chain.

Career

Where the label came from.

Started in Hyderabad, ended up in Milwaukee via a masters in Fort Collins. Front end work at first, and these days mostly the platform everything else runs on.

2022 · now

Senior Software Engineer

Northwestern Mutual · Milwaukee

Observability lead for an enterprise application rebuild spanning 4 backend services, 16 user interfaces and 2 databases. SLI design, gateway-level measurement, distributed tracing, release trains, and the CI/CD and security scanning the team ships through.

2021 · 2022

Software Engineer

Northwestern Mutual · Milwaukee

Core engineer on straight-through processing of underwriting applications. Converted the platform to a multi-case, event-driven design on AWS SQS and SNS, provisioned the messaging infrastructure in Terraform, and shipped reverse approval and decline handling end to end.

2019 · 2021

Software Developer

Egen Solutions · Chicago

Consulting engineer with enterprise clients in healthcare distribution and retail supply chain. React, Redux and Java Spring Boot on an enterprise reporting platform, D3 dashboard visualizations, and a home delivery scheduling platform.

Clients: McKesson · Sears, Innovel Solutions
2017 · 2019

MS, Computer Information Systems

Colorado State University · Fort Collins

Graduate teaching assistant for Data Visualizations along the way, teaching Tableau, Power BI, SAP Lumira and Excel to graduate students.

2014 · 2016

Software Developer

Metier Software Solutions · Hyderabad

Front end on customer-facing web applications, then across the stack: Node.js REST endpoints against SQL Server, and early container work with Docker, Ansible and Kubernetes when that was still a strange thing for a Hyderabad product team to be doing.

Toolkit

Tools I reach for.

ReactKubernetesObservability Web performance TypeScriptNode.js Java · Spring BootTerraformDocker AWSVaultGitLab CI/CD DynatraceOpenSearchPlaywright SLI · SLO designDevSecOpsMicroservices
Writing

Proof over opinion.

In the works

Working out which change actually fixed our INP

Two teams shipped in the same window and both claimed the win. A lab sweep of injected latency from 0 to 1800ms, plus field data segmented by the element each interaction landed on, settled it. Draft is in progress.

Say hello.

Milwaukee, Wisconsin. Open to good conversations about frontend, observability and platforms.