Skip to content

Open to SDE-2 roles · Backend and Full Stack

KabirMehta.

Backend engineer in Pune. I build services other teams can depend on, like the job platform at Flowgrid that runs 10,000+ jobs a day and finishes 99.9% of them.

Pune, India · IST

Snippets from queuekit, my open-source job runner

jobs a day on the platform I work on at Flowgrid
10,000+
of those jobs complete, with retries and failover
99.9%
service startup, after profiling the boot
45s to 8s
microservices built and maintained at about 99% uptime
28+

Tools I use every week

  • Java
  • Spring Boot
  • Kafka
  • Microservices
  • Azure Service Bus
  • React
  • Next.js
  • PostgreSQL
  • Redis
  • Elasticsearch
  • gRPC
  • Kubernetes

Pune, India

Now
Software engineer, Flowgrid
Also
Mock interviewer for backend roles
Focus
Backend, distributed systems, full stack
Studied
B.E. Computer Engineering

About Kabir Mehta

I am a Java full stack engineer who likes the unglamorous parts of software: retries, timeouts, startup time, and the alert that fires before a customer notices. At Flowgrid I work on the job platform. Before that I spent two years at Lumen Byte building Spring Boot microservices with gRPC, CQRS and Elasticsearch, and the React front ends to go with them.

  • Design for the retry

    Jobs fail and networks blip. The system should still finish the work, and say clearly when it cannot.

  • Measure before tuning

    45 seconds to 8 came from profiling the boot, not from guessing. The numbers decide what changes.

  • Explain what you build

    I write deep dives, take mock interviews, and review code like someone has to run it at 3 a.m.

02 · Experience

Experience. From trainee to platform engineer in four years.

Trainee to software engineer at Lumen Byte in twelve months, then the job platform team at Flowgrid.

Full resume (PDF)
  1. Software Development Engineer

    Flowgrid · Remote, Pune

    Sep 2024 - Present

    Backend engineer on the job platform: Spring Boot microservices that run large distributed workloads for supply chain customers.

    • Built the job execution service: 10,000+ jobs a day on Azure Service Bus at a 99.9% completion rate.
    • Reworked retries and error handling across services, cutting the job failure rate by 15%.
    • Cut service startup from 45s to 8s by profiling the boot and tuning configuration and the JVM.
    • Java
    • Spring Boot
    • Azure Service Bus
    • Kibana
    • Microservices
  2. Software Engineer

    Lumen Byte · Pune

    Jul 2023 - Aug 2024

    Built enterprise microservices and full stack features with Spring Boot and React.

    • Moved service-to-service calls to gRPC, cutting latency by 25%.
    • Applied CQRS and the outbox pattern for reliable events, with Kafka for streaming.
    • Interviewed Java candidates with eight or more years of experience.
    1. Associate Software EngineerApr 2023 - Jun 2023 · Noida
    2. Trainee Software EngineerJul 2022 - Mar 2023 · Noida
    • Java
    • Spring Boot
    • Kafka
    • gRPC
    • Elasticsearch
    • React
  3. B.E. Computer Engineering

    Engineering college, Pune

    2018 - 2022

03 · Selected work

Selected work. What I built, and what it changed.

  1. 01 / 04Flowgrid · 2024 - now

    Job platform

    A Spring Boot platform that runs 10,000+ business-critical jobs a day on Azure Service Bus, and keeps finishing them when things fail.

    Problem
    Thousands of jobs a day, any of which can time out, crash a worker or hit a flaky dependency.
    What I did
    Async dispatch over Service Bus, fault-tolerant scheduling with failover, retries with backoff and alerting.
    • 99.9%jobs completed
    • −85%manual fixes
    • −70%incident response
    • Java
    • Spring Boot
    • Azure Service Bus
    • Kibana

    Read the case study

    schedulerservice busworkersdoneretry · backoff 2s, 4s, 8s
  2. 02 / 04Flowgrid · Performance

    Startup time, 45s to 8s

    Every restart, CI run and cold start waited 45 seconds. Profiling the boot showed where the time went; five changes took it to 8, without touching business logic.

    • −82%startup time
    • 37ssaved per boot

    Read the write-up

    8sfrom 45s

    • Before45s
    • Scoped scanning−4s
    • Unused auto-config−3.5s
    • Lazy init−9s
    • HikariCP tuning−5s
    • Class Data Sharing−15.5s
    • After8s
  3. 03 / 04Side project · 2026

    Interview practice room

    A practice site for coding interviews: problems by topic, an in-browser editor and a judge that runs your code against test cases.

    • 400+problems
    • 48topics
    • 5languages
    • Java
    • Spring Boot
    • PostgreSQL
    • Redis
    • Next.js

    Try a problemSource code

    48topics, from arrays to union find

  4. 04 / 04Open source · 2023

    Real-time notifications

    A notification service that fans one event out to every device a person is signed in on, built on Kafka and WebSockets.

    • 2,000+active users
    • 99.8%uptime
    • Spring Boot
    • Kafka
    • WebSockets
    • Prometheus

    Source code

    post savedkafkanotifydevices

04 · Stack

The stack. Layer by layer, the way a request sees it.

  1. 01Interface

    • React
    • Next.js
    • TypeScript
    • Redux
    • React Query
    • Tailwind CSS
  2. 02Services

    • Java
    • Spring Boot
    • Spring Cloud Gateway
    • gRPC
    • WebSockets
    • OAuth 2.0
  3. 03Messaging

    • Kafka
    • Azure Service Bus
    • AWS SQS
  4. 04Data

    • PostgreSQL
    • MySQL
    • MongoDB
    • Redis
    • Elasticsearch
  5. 05Platform

    • Azure
    • AWS
    • Docker
    • Kubernetes
    • GitHub Actions
    • Nginx
  6. 06Observability

    • Prometheus
    • Grafana
    • ELK
    • Azure Monitor
  7. 07Quality

    • JUnit
    • Mockito
    • Testcontainers
    • TDD

06 · Beyond the code

Beyond the code. Interviews, mentoring and open source.

  • Mock interviews

    I take mock interviews for Java and backend roles and write up how you think, not just what you answered.

  • Interviewer

    I have interviewed Java candidates with eight or more years of experience, so I know what the other side of the table looks for.

  • Open source

    queuekit and a few smaller tools. Most of what I learn ends up in a repository or a post.

Let's build something that stays up.

Open to backend, full stack and SDE-2 roles: on-site, hybrid or remote. Email is the fastest way to reach me.

Pune, India · IST (UTC+5:30)