Sajal Halder

backend engineer — java / spring boot / microservices

Sajal Halder builds Java microservices at Rakuten scale, and now brings AI into banking.

Five years across Java and Spring Boot, from ranking, gateway and BFF services in Rakuten's e-commerce stack to AI integration as a Senior Consultant at HLB Malaysia.

  • Java
  • Spring Boot
  • Microservices
  • Apache Spark
  • Spring AI
  • RAG

Now

Senior Consultant, HLB Hong Leong Bank Malaysia

2026–present · Malaysia

Integrating AI-driven capabilities into microservice architecture across core banking systems — bringing the same reliability standards from high-throughput e-commerce infrastructure into a regulated banking environment.

  • AI integration
  • Microservices
  • Java
  • Banking

Systems built

4 production systems

Rakuten Gateway

Facade-pattern API layer aggregating upstream services into one gateway, improving search UX across Rakuten platforms.

source 1source 2source 3source 4source 5+GatewayClient
5+
upstream sources
95%+
test coverage
  • Spring Boot
  • Reactive Spring
  • SonarQube
60%
faster Spark processing on terabyte-scale daily data

Rakuten Ranking Engine

Ranking system sorting shop items by daily, weekly and monthly metrics on Dataproc Serverless — 99.5% data accuracy.

  • Java
  • Apache Spark
  • GCP
50K+
concurrent sessions on Java Virtual Threads

Rakuten BFF

Backend-for-frontend layer — replaced reactive flows with Virtual Threads for a 30% throughput gain at high demand.

  • Spring Boot
  • Virtual Threads
90%
faster reporting — 15 minutes down to under 1

ERP System

Inventory, CRM and finance modules for 150+ daily users, with transaction-safe REST APIs and RBAC.

  • Spring Boot
  • MySQL
  • Redis

Independent projects

open source on GitHub
    • Java 25
    • Spring Boot 4
    • Spring AI
    • pgvector
    • OpenAI

    A retrieval-augmented generation system where agents route each question, retrieve context and check the answer before it is returned.

    • Multi-query expansion and conversation-aware retrieval over PostgreSQL pgvector
    • Query router for intent classification, plus an evaluator that scores answer quality
    • Ingests PDF, Word, PowerPoint, Excel and HTML with structure-aware chunking; streams answers over SSE
    • TypeScript
    • Next.js 16
    • React 19
    • Supabase
    • Vercel AI SDK

    A full-stack expense tracker that reads receipt photos with AI, answers questions about spending in chat, and charts where the money goes.

    • AI reads a receipt photo, can pull several expenses out of one receipt, and asks for confirmation before saving
    • Dashboards for monthly trends, category split, payment methods and week-over-week change
    • Supabase authentication with Row Level Security, so each user only reaches their own data
  1. A Twitter-style feed that switches strategy by audience size: fan-out on write below 10K followers, fan-out on read above.

    • PostgreSQL for relationships, Cassandra for posts and timelines, Redis for feed caching
    • Asynchronous batch fan-out with TTL-based cache expiry
    • Design target of feed generation under 100 ms at p95
  2. A banking platform designed around six services (users, accounts, transactions, cards, payments and notifications) behind an API gateway.

    • Hexagonal architecture with domain, application and infrastructure layers
    • Eureka service discovery and a saga coordinator for distributed transactions
    • CI/CD stages for build, test, security scan and a SonarQube quality gate
  3. A hexagonal-architecture microservice for inventory with concurrency-safe stock adjustments.

    • Optimistic locking for concurrent updates, with pagination and validation
    • Trace IDs generated and propagated per request; Actuator health and metrics
    • Test layout mirrors the main structure, written with JUnit 5

Writing

articles
  1. Building a RAG pipeline with Spring AI and pgvector

    How a retrieval-augmented generation pipeline fits together in Spring Boot: pgvector storage, structure-aware chunking, query rewriting, routing and answer evaluation.

  2. Hybrid fan-out feed design: fan-out on write vs fan-out on read

    How a Twitter-style news feed can combine fan-out on write and fan-out on read with a 10K-follower threshold, using PostgreSQL, Cassandra and Redis.

  3. Building an AI receipt scanner with Next.js, the Vercel AI SDK and Supabase

    How Expense Tracker AI turns a receipt photo into saved expenses with Next.js 16, the Vercel AI SDK and Supabase Row Level Security, and why a person confirms first.

All articles

Experience

  1. 2026 — present

    Senior Consultant

    HLB Hong Leong Bank, Malaysia
    • Integrating AI capabilities into microservice architecture within core banking systems
  2. Mar 2023 — May 2026

    Software Engineer

    Bangladesh Japan IT Ltd. (BJIT) · Dhaka
    • Contributed to microservices used by Rakuten that handle billions of requests daily
    • Collaborated with international teams and stakeholders to align on business objectives
    • Participated in peer code reviews, giving feedback to keep code quality high
  3. 2017 — 2021

    B.Tech, Computer Science & Engineering

    Lovely Professional University · Punjab, India

Skills

grouped by layer

AI

  • Spring AI
  • RAG pipelines
  • AI agents
  • LLM integration
  • Vector search (pgvector)

Services

  • Java
  • Spring Boot
  • Microservices
  • Reactive Spring
  • Virtual Threads
  • REST API design
  • Hexagonal architecture
  • Saga pattern
  • JPA / Hibernate

Data

  • PostgreSQL
  • MySQL
  • Redis
  • Cassandra
  • Apache Spark

Platform

  • Docker
  • Kubernetes
  • Jenkins
  • GitHub Actions
  • GCP (Dataproc)
  • Prometheus
  • Grafana

Quality

  • JUnit 5
  • SonarQube
  • Integration testing
  • Code review

Front end & mobile

  • TypeScript
  • Next.js
  • Flutter
  • Dart

About Sajal Halder

Sajal Halder is a Java and Spring Boot engineer with five years of experience building backend systems. Sajal began by building an ERP platform for 150+ daily users, then joined BJIT to work on microservices used by Rakuten, including ranking, gateway and backend-for-frontend services.

Since 2026, Sajal has been a Senior Consultant at HLB Hong Leong Bank Malaysia, integrating AI into microservice architecture. Sajal holds a B.Tech in Computer Science and Engineering from Lovely Professional University.