Software Development Engineer @ Birdeye

Backend Engineer — Designing For Scale & Speed

Software Engineer with 4.5+ years of experience specializing in building high-throughput distributed backend infrastructure. Architect of event-driven data pipelines handling 50M+ daily operations at 99.9% uptime, with custom fault-tolerant systems for high-throughput workloads.
View ProjectsContact MeGitHub
arsh.ts
// Arsh Radhanpura

import { BackendEngineer, DistributedSystems,
 AIInfrastructure, } from "arsh-core-engineering";

export class Engineer extends BackendEngineer {
  public readonly name = "Arsh Radhanpura";
  public readonly title = "Software Development Engineer @ Birdeye";
  public readonly experience = "4.5+ Years";

  public readonly focusAreas = [
    "High-Performance Distributed Systems",
    "AI & LLM Infrastructure",
    "Event-Driven Architectures",
  ];

  /**
   * Building systems that scale, recover, and stay reliable.
   */
  public async engineer(system: DistributedSystems) {
    return system
      .scale({ operations: "50M+ daily" })
      .reliability({ uptime: "99.9%" })
      .optimize({ performance: "High-Throughput" })
      .addIntelligence({ stack: "AI / LLMs" })
      .ship({ environment: "Production" });
  }
}
scroll

Experience

2025 — Present · 1 yr 8 mo● ACTIVE
Birdeye logo

Birdeye

Software Development Engineer II

Palo Alto, California · Remote

Architecting distributed systems and AI-powered infrastructure across web crawling, database scalability, high-throughput services, and production reliability.

  • Built an AI-powered self-healing web-crawling framework combining deterministic heuristics, confidence-scored LLM fallbacks, human-in-the-loop validation and circuit-breaker resilience.
  • Engineered a low-latency global transaction sharding architecture using PostgreSQL partitioning, Foreign Data Wrappers (FDW), and AWS RDS Proxy.
  • Engineered LLM-driven adaptive crawling using LangChain and OpenAI agents to dynamically select HTML, REST, and GraphQL extraction strategies.
  • Separated an 800 GB production database from a shared database with zero downtime and zero data loss, improving scalability and reliability.
  • Architected and executed a zero-downtime MongoDB v3 → v7 migration across distributed microservices with phased rollout and rollback automation.
  • Improved P95 latency of three high-traffic Listing APIs by 40% through MongoDB aggregation optimization, query tuning, profiling, and distributed caching.
  • Designed an end-to-end observability and alerting framework with automated Slack/email incident notifications, reducing production MTTD.
  • 40% reduction in P95 latency across three high-traffic services.
  • Separated an 800 GB production database with zero downtime and data loss, removing a critical scalability constraint.
  • Completed a zero-downtime MongoDB v3 → v7 migration across distributed microservices.
  • Enabled cross-region transaction scaling using PostgreSQL partitioning, FDW, and RDS Proxy.
  • Eliminated manual crawler maintenance by 65% through an AI-powered self-healing recovery system.
  • Reduced production MTTD with end-to-end tracing and automated incident alerting.
  • Delivered a scalable reporting platform generating $100K+ in annual revenue.
  • Standardized CI/CD infrastructure across teams using Jenkins and Kubernetes-based dynamic agents.
Node.jsTypeScriptPuppeteerLangChainOpenAIKafkaRedisMySQLMongoDBPostgreSQLKubernetesJenkins
2022 — 2025 · 3 yrs
Bajaj Finserv Health logo

Bajaj Finserv Health

Software Development Engineer I → Software Development Engineer II

Pune, India

Progressed from SDE 1 to SDE 2, architecting backend systems across payments, Billing, authentication, real-time notifications, reporting, and performance engineering.

  • Architected payment orchestration and partial payments systems supporting payouts, refunds, split payments, settlements, and subvention workflows.
  • Designed a multi-tenant OAuth authentication platform with JWT dual-token flows, session management, token revocation, and distributed refresh deduplication.
  • Built an in-house real-time WebSockets notification service using Socket.IO, and Kafka.
  • Migrated synchronous reporting workflows to event-driven architectures, enabling asynchronous and independently scalable processing.
  • Built and published caching Artifacts adopted across multiple large scale microservices.
  • Scaled Puppeteer-based PDF generation with concurrent Chromium instances.
  • Established and streamlined CI/CD infrastructure across automated builds and deployment workflows.
  • Cut production costs by 18% with the in-house WebSocket notification service
  • Reduced transaction costs by ~2% and increased online payment adoption by 33%
  • Improved P90 API latency times by 38% and reduced DB load by 42.48% and QPS by 65.36%
  • Reduced average CI/CD pipeline time by over 50% through dependency caching and parallelization
  • Reduced average cache retrieval latency by 40% with a shared Redis library
  • Increased PDF generation throughput by 21% and cut failure rates to zero
  • Reduced DB load by 45% migrating reporting workflows to asynchronous Service Bus
  • Increased PDF generation throughput by 21% using concurrent Chromium instances.
TypeScriptNode.jsNestJSApache KafkaAzure Service BusRedisMongoDBElasticsearchSocket.IOPuppeteerAzureArgoCD

Engineering Impact

Measured outcomes from production systems, pulled directly from work history.

38%

P90 API Latency Reduced

Distributed system performance tuning

43%

Database Load Reduced

Database load and Query per second (QPS)

40%

PDF generation concurrent throughput

PDF generation concurrency of Chromium instances.

33%

Online Payment Adoption Increased

Payment orchestration system

18%

Production Costs Reduced

In-house WebSocket notification service

50%+

Average CI/CD Time Reduced

Dependency caching & pipeline parallelization

About

// who-am-i.md

I'm a Software Engineer with 4.5+ years of experience building scalable, reliable backend systems and distributed infrastructure. I currently work at Birdeye, where I design and build backend services, event-driven systems, and data-intensive infrastructure for a B2B Agentic AI platform. My work spans system design, asynchronous processing, performance optimization, observability, and building systems that remain reliable under high throughput.
Before Birdeye, I spent three years at Bajaj Finserv Health, owning backend systems across payments, authentication, notifications, reporting, and core platform services. I worked on systems where reliability and correctness were critical, from designing end-to-end business workflows to improving performance and handling failures across distributed services. This experience shaped how I approach backend engineering: thinking carefully about scalability, consistency, failure modes, and the operational behavior of systems in production.
The engineering problems I enjoy most sit at the intersection of scale and reliability: keeping a system correct when messages can be duplicated, retried, or delivered out of order; squeezing latency out of a hot path without sacrificing correctness; and designing infrastructure that degrades predictably instead of failing silently.

Technical Skills

Technologies used in production systems.

Distributed Systems & Messaging

  • Kafka
  • RabbitMQ
  • Azure Service Bus
  • Socket.IO

Infrastructure & Observability

  • Kubernetes
  • Docker
  • AWS
  • Jenkins
  • ArgoCD
  • Grafana
  • Prometheus
  • ELK Stack

AI & Agentic Systems

  • LangChain
  • OpenAI APIs

Backend & Frameworks

  • Node.js
  • NestJS
  • Express.js
  • Spring Boot
  • Golang

Databases & Caching

  • MongoDB
  • PostgreSQL
  • MySQL
  • Redis
  • Elasticsearch

Languages

  • TypeScript
  • JavaScript
  • Python
  • Java
  • Go

Crawling & Automation

  • Puppeteer
  • Playwright
  • Chrome DevTools

Projects

Production systems built to handle scale, failure, and change.

33% Payment Adoption ↑

End-to-end healthcare payments system integrating Razorpay, Salesforce, appointment systems, and doctor payout workflows.

70%+ Fewer Manual Fixes

An AI-assisted self-healing agent that recovers crawlers from broken selectors when provider websites change their DOM structure, without manual engineering intervention.

0% Payout Failures

Eliminated Payout inconsistencies where ~0.3% of transactions left doctors unpaid despite successful patient charges.

10K+ Reviews Handled

End-to-end pipeline automating review ingestion, sentiment analysis, AI-generated responses, and operational insights across platforms like Google and Facebook.

No-Code Source Demos

A workflow letting the sales team demo unsupported review sources live during enterprise pitches, without upfront engineering work.

10K+ Concurrent Sessions

OAuth-inspired authentication and authorization microservice for a multi-tenant doctor platform, balancing low-latency auth with instant revocation and multi-device sessions.

Contact

contact.sh

$ connect --with "Arsh Radhanpura"

$ status

OPEN_FOR_OPPORTUNITIES