Available for new opportunities

Soham Deb Majumder

Backend Engineer  ·  exploring full-stack & AI

I build scalable distributed systems, event-driven architectures, and cloud-native solutions. 2 years deep in backend, now expanding into full-stack and AI engineering.

View Experience ↓ GitHub →

Roots in backend, reaching everywhere

foundation
2 years building distributed systems at fintech and AI product companies — Kafka pipelines, microservices, cloud infra that handles real production load.
Exploring full stack and AI. Building UIs, learning the frontend side of things, and integrating LLMs into backend systems. Still early, but doing it.

What I work with

⚙️ Languages
Go TypeScript JavaScript Java Python
🖥️ Backend
Node.js Gin FastAPI Spring Boot REST APIs Async Processing
🗄️ Data & Messaging
Kafka Redis PostgreSQL MongoDB BullMQ
☁️ Cloud & DevOps
AWS Lambda S3 EC2 Kubernetes Docker GitHub Actions
🤖 AI Integration
OpenAI APIs LangChain Prompt Engineering Cursor Copilot
🔍 Observability & Concepts
OpenTelemetry Kibana Microservices Event-driven Distributed Systems

Where I've worked

SDE I
May 2025 – Feb 2026
Smallcase
Hybrid · India · Fintech
  • Integrated RPD feature across web and mobile user journeys, reducing user drop-offs by 20%
  • Built real-time alerting system integrating multiple data sources, cutting incident detection time by 60%
  • Developed Kafka-based retry and failure handling system achieving 99.99% message delivery reliability
  • Automated data ingestion pipelines using AWS Lambda, reducing manual processing effort by 70%
  • Optimised Docker builds from hours to 10 minutes using multi-stage builds
  • Resolved 10+ critical production issues, reducing order failures by 10%
Software Engineer
Jan 2024 – Apr 2025
Tapfinity Private Limited
Remote · India · AI Product
  • Built AI-driven onboarding system increasing user conversion by 40%
  • Designed AI-driven coaching system with event-driven workflows, improving engagement & retention by 25%
  • Designed event-driven workflows using Kafka and BullMQ, improving processing efficiency by 40%
  • Reduced infrastructure costs by 30% through cloud migration and microservices architecture
  • Improved observability and CI/CD pipelines, reducing MTTR by 30%
  • Enhanced search relevance by 50% using indexing and tokenisation strategies

Things I've built

e-ticketing-system
High-performance ticket booking system built with Java Spring Boot, designed for massive concurrent load with zero overselling and full seat consistency.
50K concurrent users P95 < 400ms 100% seat consistency Zero overselling
Java Spring Boot Kafka Redis
View on GitHub →
rag-pipeline-with-eval-harness
Production-style RAG pipeline on local LLMs with a custom eval harness measuring faithfulness, context recall, and answer relevancy. Uses HyDE query rewriting and hybrid retrieval with BM25 + dense search fused via RRF.
Local LLMs 3 eval metrics Hybrid retrieval
Python FastAPI Qdrant Ollama Docker
View on GitHub →
api-gateway-go
Lightweight API gateway in Go with config-driven routing, reverse proxying, JWT auth, Redis-backed rate limiting, Prometheus metrics, circuit breaker, and hot-reloadable config — no restart needed.
Hot config reload Circuit breaker Redis rate limiting
Go Redis JWT Prometheus Docker
View on GitHub →

Milestones

🏆
Winners — NASA SpaceApps Challenge 2022
Global Impact Award winners at the NASA SpaceApps Challenge 2022 — one of the world's largest hackathons with participants from 160+ countries. Built a solution addressing real-world problems using space data and open-source technologies.
🔗 View certificate →

Let's build something

I'm open to backend roles, full-stack opportunities, and AI engineering positions. If you're building something ambitious that involves scale, I want to hear about it.

✉ sohamdebmajumder@gmail.com in linkedin.com/in/soham-deb-majumder ⌥ github.com/samwilson0745 ☎ +91 80170 98639

// quick bio in JSON

$ cat soham.json
{
  "role": "Backend Engineer",
  "exp_years": 2,
  "core": ["Go", "Node.js", "Kafka", "AWS", "MongoDB"],
  "exploring": ["full-stack", "AI engineering"],
  "open_to": true
}