Ihsan Mirani Rumi

Platform Engineering Team · US-Bangla Group · Dhaka, Bangladesh

• Managed infrastructure automation and GitOps pipelines across a 20+ microservice architecture for CarryBee, a large-scale courier and logistics platform serving 100K+ daily users.
• Implemented backend functionality for a major transport management system feature, building single, bulk, and batch Go/gRPC and REST APIs consumed by operations and agent-facing services across multiple microservices; shipped to staging and resolved issues surfaced during testing and rollout.
• Collaborated on developing migration automation and validation workflows for the production environment’s migration from GCP to AWS EC2 covering 1.5B+ records, contributing to a 65%+ reduction in migration time after process improvements.
• Assisted in R&D and migration of production workloads to AWS ARM (Graviton) instances, leveraging Go’s strong ARM performance to cut infrastructure costs by up to 20%.
• Participated in migrating the staging environment from a cloud provider to self-hosted infrastructure, helping reduce infrastructure costs by over 70%.
• Rebuilt CarryBee’s staging environment from scratch after its transfer from TechnoNext to the Platform team, migrating version control and all service code and standing up deployments and observability, including custom Grafana dashboards, end to end.
• Collaborated in diagnosing and resolving live production issues, including performance bottlenecks and service reliability improvements.

DevOps Team · US-Bangla Group · Dhaka, Bangladesh

• Provisioned infrastructure and CI/CD pipelines across multi-environment systems (CarryBee, ERP, Lunch Management System) serving up to 100K+ daily users, managing automated backups and production deployments.

A personal platform-engineering and SRE lab built entirely on free tiers, run like production.

Migration automation and validation workflows for a production move covering 1.5B+ records, contributing to a 65%+ reduction in migration time.

Migrating the staging environment from a cloud provider to self-hosted infrastructure, helping reduce infrastructure costs by over 70%.

R&D and migration of production workloads to AWS ARM (Graviton) instances, cutting infrastructure costs by up to 20%.

Compared emotion-recognition models, from action-unit-based models to deep CNNs, across Western (KDEF) and Bengali datasets.

Shahjalal University of Science and Technology (SUST)

CGPA: 3.48 / 4.00 · Thesis: Comparative Analysis of Emotional Expressions — KDEF vs. Bengali Dataset

Notre Dame College

GPA: 5.00