mahesh babu narne · senior software engineer ·

I build distributed systems that stay fast, observable, and quiet at 3 a.m.

Four-plus years building scalable distributed backend systems in Java, Python, and TypeScript across communications and healthcare platforms. Currently a Senior Software Engineer at Apple, where the platform I work on delivers email, SMS, and push to millions of daily users without raising its voice.

tail -f prod ›p99 latency −35% after routing fix

trace · work

trace_id: career · 2 spans · 0 errors2021 → now
202120222023202420252026
  • Built and operated high scale Java backend services for a consumer communications platform delivering email, SMS, and push to millions of daily users owned end to end through deployment, monitoring, and on-call.
  • Designed REST and GraphQL APIs consumed by multiple product teams, investing in schema stability and backward compatibility so downstream teams ship independently.
  • Built React and TypeScript interfaces for internal tooling and consumer facing notification preferences, working closely with product and design.
  • Developed Kafka event pipelines for real time delivery and engagement tracking, with partition strategy and consumer group isolation that keeps latency predictable under peak load.
  • Integrated ML outputs and LLM powered features into production Java services, handling inference latency and partial failures gracefully without degrading the core experience.
  • Built observability on Prometheus, Grafana, and OpenTelemetry structured around SLOs and real failure categories actionable signals, not metric noise.
  • Found and fixed a message routing bottleneck that cut p99 latency 35% and improved throughput 50%, after profiling revealed the actual slow path.
  • Mentored two engineers through complex platform features, helping them reason through failure modes before production did.

metrics · impact

−0%

p99 latency on the message routing path, after profiling the real slow path

+0%

throughput on that same path once the bottleneck was gone

−0%

mean time to detection, via SLO driven alerting on real failure categories

−0%

mean time to resolution on a production distributed system

−0%

on call response time, from runbooks grounded in how the system actually fails

3 teams

adopted the backend platform within two quarters no structural rewrites

stack · by layer

interface

React · Angular · TypeScript · component architecture

services

Java · Spring Boot · Python · Kotlin · C# · REST · GraphQL · gRPC

data

PostgreSQL · MySQL · MongoDB · DynamoDB · Redis · Elasticsearch

messaging

Apache Kafka · SQS · event-driven pipelines · async processing

infrastructure

AWS · Azure · GCP · Kubernetes · Docker · Terraform · CI/CD

observability

Prometheus · Grafana · OpenTelemetry · distributed tracing · SLOs

ai / automation

LLM integration · agentic workflows · ML feature integration · Claude Code

credentials

certifications

Microsoft Azure Solutions Architectexpert
Microsoft DevOps Engineerexpert
AWS Certified Solutions Architectassociate
Certified Kubernetes Administratorcka

education

M.S. — University of Central Missouricomputer science
B.Tech — GITAM Universitycomputer science

contact · open

Building something that needs to scale without drama? Let’s talk.

open to relocation · listening on :new-roles (actively interviewing)