Backend systems, platform tooling, APIs
Hi, I'm Debashis Mitra,
software engineerfrom Kolkata, India
I design and implement solutions to problems in society into application systems based on website and mobile platforms.
What I Work On
Not a skills list. These are the responsibilities I take on.
Backend Systems
Designing and operating REST APIs with Spring Boot, applying DTO validation, security chains, OAuth2, and JWT. Building services that handle real traffic.
Distributed & Data
Event-driven architecture with Kafka, multi-database design (relational + document + cache), and systems that stay consistent under load.
Cloud & Operations
Deploying and operating services on Azure VMs with Docker, Nginx, SSL termination, secrets management, and uptime monitoring.
Developer Tooling
Building SDKs, CLI tools, and internal platforms that other developers rely on. Integration-first thinking means less friction and more reliability.
Upblit
Problem
Development teams end up stitching together separate tools for observability and secrets management, each with its own auth model, its own dashboard, and no shared context. Upblit unifies both into one multi-tenant platform with a single source of truth.
What it does
Teams instrument applications with lightweight SDKs (Express, Python, Java, Go) that stream logs, telemetry, and traces into a Kafka ingestion pipeline, scoped by tenant, project, organisation, and application. The same platform manages client-facing secrets through a purpose-built Lua-based manager, while internal infrastructure credentials stay isolated in HashiCorp Vault.
Architecture
Engineering Challenges Solved
Tech Used
Other Systems I've Built
The flagship carries weight. These show consistency.
A multitenant distributed job processing platform built with Go, PostgreSQL, and Redis, featuring priority queues, retries with exponential backoff, scheduling, webhooks, RBAC, and an operations dashboard.
Practical Experience
Framed by what was delivered, not job titles.
Backend Engineer - Freelance
2024 - PresentDesigning and delivering backend systems for client products, from initial architecture through production operations.
- Integrated transactional email infrastructure with SPF/DKIM and queue-backed delivery
- Deployed production APIs on cloud VMs with zero-downtime rollout strategy
- Built API key issuance and scoped access control for multi-tenant clients
- Set up monitoring and alerting for client-facing services
OpenTelemetry Java SDK - Contributor
Open Source ContributionContributed reliability, lifecycle management, and compatibility improvements to a widely used observability ecosystem.
- Fixed OkHttp client mTLS handling when using the platform's default trust store
- Added shutdown lifecycle support for resource-owning Samplers, allowing executors and connections to be released during SDK shutdown
- Implemented SDK self-observability version gating for PeriodicMetricReader
- Worked within an established Java codebase, following existing design conventions and project contribution standards
Shipd - Project Olympus
Mar 2026 - PresentContribute to real-world software engineering challenges used to train and evaluate AI coding models.
- Translate practical software engineering problems into clear, self-contained, reproducible coding challenges
- Design behavior-focused tests to validate expected functionality and catch incorrect implementations
- Build isolated Docker-based environments with the required dependencies and execution setup
- Use the Olympus CLI, prechecks, and verification workflows to validate challenge quality and reproducibility
- Identify edge cases, ambiguous requirements, and environment issues that could affect evaluation reliability
- Focus on making engineering tasks realistic, deterministic, and technically meaningful
Engineering Notes
Writing proves understanding. Understanding reduces hiring risk.
Codily's Redis ZSET Delayed Queue
How I built delayed job scheduling using Redis Sorted Sets, Unix epoch nanosecond scores, and a scheduler goroutine that atomically promotes ready jobs with exponential backoff support.
ReadWhat I Learned Deploying 8 Services Across GCP, AWS, Azure, and Cloudflare
An honest reflection on deploying Upblit and Codily across multiple cloud providers, covering networking challenges, secrets management, infrastructure costs, and operational lessons learned as a solo developer.
ReadIdempotency Keys in Distributed Job Systems
Why async systems need idempotency, how Codily deduplicates requests with the Idempotency-Key header, and how this prevents duplicate emails, webhooks, and background jobs.
ReadOpen Source & Community Work
Hacktoberfest Maintainer
Reviewed pull requests, triaged issues, mentored first-time contributors, and kept community submissions production-minded.
Community Building
Grew a project to 1k users organically with contributor onboarding, clear guidelines, and steady issue triage.
Issues & Code Review
Resolved 60+ issues and reviewed public patches while maintaining backward compatibility across a public API.
Technical Stack
Tools do not differentiate the work. Responsibilities do.
Get in Touch
Open to backend and platform engineering roles and internships.