Skip to content

Backend systems with a quiet interface and production-grade AI.

I build distributed systems, microservices, and production AI/ML workflows shaped by strong system design.

View Work
GitHub
available for workPune, IN · UTC+5:30 · 2yrs in production
~/mohammadumar — system.statusoperational
stack
Spring Boot · PostgreSQL · Redis · LLM systems
focus
Distributed systems · AI/ML workflows
location
Pune, IN · UTC+5:30
status
available_for_work

GitHub velocity · 2026 quarters

contributionsPRs
Q1Q2Q3Q4

Languages · owned repos

  • Java 53.5%
  • TypeScript 25%
  • Other 7.5%
  • JavaScript 7.2%
  • Go 4.2%
  • Python 2.5%
contributions 1,123 · PRs 81 · 2026

01 AI systems shipped

4+

RAG, agents, and production LLM features

02 GitHub contributions all-time

1,297+

public projects and production delivery

03 years in production

2

fintech · healthcare · govtech · artificial intelligence · enterprise saas

04 products shipped

10

owned end to end or contributed in production

01 — selected work

Things I own end to end.

01 · AI / Career Tech · 2025

ResumeAgent

A 5-agent AI pipeline that turns one master resume into job-specific, ATS-optimized documents.

Production-grade multi-agent system in Java 21 / Spring Boot: parser, JD analyzer, semantic matcher, rewriter, and ATS optimizer orchestrated over structured JSON, with a model-agnostic LLM layer and deterministic FreeMarker→DOCX rendering.

→ AI writes the content — templates guarantee the layout. Zero formatting hallucinations.

Java 21Spring AIPostgreSQL JSONBMulti-Agent
Read case study

02 · Developer Tooling · 2025

GitPilot

A cross-platform AI Git CLI — conventional commits, semantic grouping, and PR summaries with a human always in the loop.

Go-based CLI that reviews local changes, generates conventional commit messages, splits work into semantic commit groups, and writes PR descriptions — through a provider-agnostic LLM architecture with approval gates before every commit and push.

→ Zero plaintext credentials — API keys live in the OS keychain.

GoGroqOS KeychainGoReleaser
Read case study

03 · Automotive / Dealership · 2025

Carventory

A three-app dealership platform covering the full workflow — inventory to invoice to profit/loss.

Production monorepo with a Spring Boot REST API, a React + TypeScript staff dashboard, and a Next.js public marketplace — inquiry-to-vehicle matching, bookings, branded PDF invoices, and real-time financial analytics under JWT multi-role RBAC.

→ One monorepo, three apps, the entire dealership workflow.

Java 21Spring Boot 3.4React 18Next.js 15
Read case study

04 · Healthcare / Diagnostics · 2024

PathLab

An enterprise LIMS that takes a pathology sample from booking to signed PDF report.

Laboratory Information Management System automating the full pathology workflow — online booking, multi-status sample lifecycle tracking, parametric result entry, and secure PDF report delivery — under 4-role RBAC with Caffeine-cached hot paths.

→ Six-state sample lifecycle, four roles, zero paper.

Spring BootReactPostgreSQL4-role RBAC
Read case study

05 · Browser Extension · 2025

DWUI New Tab

A glassmorphism new-tab experience with a WebGL aurora — live on the Chrome Web Store.

Chrome extension replacing the default new-tab page with a WebGL-powered animated aurora, glass panels, smart search over the Chrome API, and a drag-and-drop shortcuts grid — built with React 19 and OGL.

→ Published on the Chrome Web Store — a daily-driver for real users.

React 19OGL / WebGLVite 7Chrome APIs
Read case study

02 — live activity

The work, in real time.

~/activity $ tail -f github.log

live · proof of work· synced just now

contributions · all-time

1,297

2026 · contributions

1,123

2026 · pull requests

81

current streak

2days

~/contributions — last 27 weeks
MarAprMayJunJulAugMonWedFri
856 contributions in window
~/pull-requests — recent
~/repos — recently pushed
~/languages — owned repos

Languages · owned repos

  • Java 53.5%
  • TypeScript 25%
  • Other 7.5%
  • JavaScript 7.2%
  • Go 4.2%
  • Python 2.5%

03 — professional work

Production systems I contribute to at Data Innovation Technologies.

01

Examination System

EdTech

Self-hosted secure MCQ examination platform — full-screen enforcement, tab-switch detection, one-attempt integrity, timed auto-submit.

▲ 10,000+ concurrent users

Next.js 16React 19Prisma 7PostgreSQLZod
02

SimpleAccounts UAE

FinTech

Comprehensive accounting software for UAE businesses — invoicing, banking, VAT compliance, payroll, and financial reporting.

Spring BootReact 18PostgreSQLLiquibasek6
03

TherapyCRM

Healthcare

Multi-tenant, HIPAA-compliant therapy practice management SaaS — scheduling, treatment plans, billing, and a JWT mobile API.

React 18Node.jsExpressDrizzle ORMStripe
04

Doctor Website

Healthcare

Premium public site and custom CMS for a metabolic health clinic — architecture designed and built end to end.

Next.js 16React 19Prisma 7PostgreSQLGroq SDK
05

Corporate App / SIR

GovTech

Civic complaint management plus voter-record mapping across election years — Jaro-Winkler fuzzy matching and a WhatsApp self-service bot.

▲ 4 user roles in production

Java 17Spring Boot 3.2PostgreSQLFlywayReact 19

04 — experience

Two years of production, one promotion earned.

  1. f558752 · Apr 2025 — Present

    Software Engineer (Java)

    ↗ promoted in 8 months

    Pune, India

    Java 17Spring Boot 3.xSpring AIPostgreSQLRedisDocker
    Scaled a secure examination platform to 10,000+ concurrent users by designing a layered caching architecture — Redis distributed cache, PgBouncer connection pooling, and client-side caching — keeping the database out of the hot path.
    Cut API response times by 60% by profiling slow endpoints, optimizing PostgreSQL queries, adding strategic indexes, and restructuring data-access pipelines for read-heavy workloads.
    Delivered a govtech civic-complaint and voter-mapping platform serving citizens across 4 user roles — Spring Boot 3.2, fuzzy record matching across electoral rolls, and a WhatsApp self-service bot.
    Shipped AI-driven features to production across Java and Next.js services — Spring AI, Groq SDK, context-managed prompts, and structured outputs — including an end-to-end corporator helpline system.
    Owned zero-downtime deployment of the systems I built with Docker, Nginx load-balanced reverse proxies, and GitHub Actions CI/CD.

  2. c040c25 · Aug 2024 — Apr 2025

    Software Engineer Intern (Java Backend)

    Pune, India

    JavaSpring BootSpring SecurityPostgreSQLLiquibaseJUnit 5
    Shipped major production features for SimpleAccounts.io (fintech accounting) and TherapyCRM.io (healthcare CRM) — RESTful APIs and invoicing/reporting modules across the full SDLC.
    Secured multiple production APIs with JWT authentication (access + refresh token rotation), RBAC, and input validation via Spring Security.
    Preserved data integrity through evolving requirements with relational schema design and version-controlled Liquibase/Flyway migrations.
    Guaranteed consistent dev-to-production behavior with JUnit unit/integration tests, dynamic FreeMarker email template systems, and Dockerized services in Agile sprints.

05 — capabilities

The stack I run in production.

umar@prod:~ — zsh

~/mohammadumar $ cat capabilities.yaml

backend/:

Java 8–21Spring Boot 3.xSpring SecuritySpring Data JPASpring AIHibernateMaven

systems/:

MicroservicesApache KafkaRedisCaffeineNginx Load BalancingPgBouncerRate LimitingHorizontal Scaling

ai-llm/:

Multi-Agent PipelinesRAGContext ManagementStructured OutputsMCP ServersOpenAI · Claude · Groq · Ollama · OpenRouter

data/:

PostgreSQL + JSONBRedisMySQLFlywayLiquibaseQuery Optimization & Indexing

frontend/:

React 18/19Next.js 15/16TypeScriptTailwind CSS

ops/:

DockerGitHub ActionsNginxLinuxAWSJUnit 5SonarQubek6 Load Testing
capabilities.yaml · 6 groups · 39 entries · read-only

06 — about

Systems first. Always.

I'm a software engineer at Data Innovation Technologies in Pune, where I earned a promotion from intern to full-time engineer in eight months by consistently shipping production releases across fintech, healthcare, and govtech.

My work lives where systems meet scale: layered caching architectures that keep databases out of the hot path, PostgreSQL tuned for read-heavy workloads, and services deployed behind load-balanced, zero-downtime pipelines.

Lately I ship LLM features that actually reach production — multi-agent pipelines, RAG, and model-agnostic orchestration — built on the belief that AI generates the content, but systems guarantee the outcome.

AI generates the content. Systems guarantee the outcome.
~/facts.json
location
Pune, IN
employer
Data Innovation Technologies
fintech · healthcare · govtech
education
B.Sc. in Computer Science
Savitribai Phule Pune University · 2026 · CGPA 8.58
daily drivers
Java · Spring Boot · PostgreSQL
~/me.png
Mohammad Umar Shaikh

06 — contact

Let's build something that scales.

available_for_work · replies within a day
mohammadumar.dev@gmail.com✓ copied

~/mohammadumar $ ls ~/channels