Open to backend & ML engineering roles

Md Mahabubul Hasan

Software Engineer · fintech · AI · IoT

I build high-availability backend systems and cloud-native infrastructure — across fintech, AI, and IoT.

Copenhagen, Denmark

Selected work

Projects I'm proud of

A real-time database library, a carbon-aware cloud optimizer, an AI-native VCS, and a production payment gateway. The deep ones, not the demos.

LiveSQL

Open source · npm

GitHub npm

Real-time sync for PostgreSQL. When a row changes, every connected client sees it instantly — no polling, no manual invalidation, no infrastructure changes.

37,872/sevents at p95 96ms · 1,000 clients · 0 failures

WAL-based CDC logical replication with offset-based backfill and event replay on reconnect.
Multi-package SDK with React, Vue, Svelte and vanilla-JS bindings; JWT auth, row-level permissions, filtered column access.
Backpressure handling, WAL-slot monitoring and failover recovery — covered by 163 unit + 23 chaos tests.

TypeScript
PostgreSQL
WebSockets
React
Vue
Svelte

Canopy Cloud

Open source

GitHub

A budget- and carbon-aware infrastructure architect. Canopy scans your cloud and produces a unified EcoWeight score that blends cost and carbon for every workload.

30–35%of cloud spend targeted as overprovisioning waste

EcoWeight scoring, idle detection, right-sizing and greener-region migration across 48 AWS + GCP regions.
CARL scheduler defers workloads by grid carbon intensity; shift-left `canopy plan` for Terraform & Pulumi.
MCP servers expose billing/carbon/Slack/GitHub tools to LLM hosts, with interactive, Slack or GitHub approval flows.

Python
FastAPI
AWS
GCP
MCP
Agentic AI

gpp (git++)

Open source · Rust

GitHub

An AI-native version control system. gpp replaces Git's commit-centric model with a continuous-capture architecture built for a world where AI agents are first-class contributors.

9 / 9architecture phases implemented

Content-addressed store (BLAKE3 + zstd), tree-sitter semantic diffs, and a full Git import/export bridge.
Encrypted knowledge graph, Noise_XX P2P sync, policy engine, per-changeset cost tracking, and a ratatui TUI.
Multi-crate Rust workspace — all 9 phases implemented, 133 tests passing, clippy & fmt clean.

Rust
tree-sitter
BLAKE3
MCP
P2P

Moneybag Payment Gateway

Fingerprint IT · Work

A Bangladesh Bank PSO-compliant payment gateway built from the ground up — settlement, reconciliation, and a multi-tenant merchant platform.

3 DCshigh-availability Patroni cluster with auto-failover

Automated settlement engine: configurable rules, EMI installments, recurring invoicing, subscription billing, payment links.
Bank reconciliation, dispute resolution, refund/hold workflows and end-to-end audit trails for regulatory review.
PostgreSQL Patroni cluster across 3 data centers with automatic failover; full Prometheus/Grafana/Loki/Sentry observability.

Python
FastAPI
Patroni
Celery
Redis
Docker

See more on GitHub

Experience

Where I've worked

3+ years across payments, AI platforms, and computer-vision systems.

Software Engineer · Fingerprint Information Technology Ltd
Dec 2024 – Aug 2026
Dhaka, Bangladesh
- Primary backend engineer on the Moneybag payment gateway, built from scratch and licensed under Bangladesh Bank (central bank) rules; 30+ merchants onboarded.
- Integrated 11 payment processors: 5 mobile wallets (bKash, Nagad, Rocket, Upay, Tap) and 6 bank and card acquirers covering Visa, Mastercard and Amex.
- Built HMAC-signed merchant webhooks with retries, safeguards so no payment is ever processed twice, and a circuit breaker around processor status checks.
- Ran PostgreSQL on a three-datacentre Patroni cluster with streaming replication, monitored with Prometheus, Grafana, Loki and Sentry; moved deploys to versioned container images.
- Python
- FastAPI
- PostgreSQL
- Patroni
- Celery
- Redis
- Docker
- Grafana
Software Engineer (Part-time, Remote) · Yildiz Consultancy
Aug 2025 – Aug 2026
Remote, United States
- Sole author of a rebuilt multi-tenant media localisation platform for AI voice, translation and subtitles in 13 languages (ElevenLabs, DeepL, LanguageTool, Zemberek).
- Migrated the legacy production data into the new schema on Amazon RDS: 326 projects and 50,691 words.
- Built the GitHub Actions release pipeline for My Smart Homes, a white-label Home Assistant distribution, publishing multi-architecture container images and OS images.
- Moved the fork onto Home Assistant Core 2026.4 and OS 17.1, carrying 69 consumer and 14 enterprise patches, and verified the new pipeline against the old one.
- Python
- FastAPI
- React
- TypeScript
- Home Assistant
- GitHub Actions
- AWS
Software Engineer · Uhurutek Solutions Ltd
Sept 2023 – Nov 2024
Dhaka, Bangladesh
- Cut customer support response time by 50% with RAG chatbots built on LangChain and the OpenAI API.
- Reached 92% retrieval accuracy on complex tax queries through prompt engineering, context tuning and a custom PDF, Word and web ingestion pipeline.
- Deployed the system on-premise for a government tax agency, with no data leaving its servers.
- Put an API gateway (Keycloak auth, APISIX rate limiting) in front of the services and migrated legacy infrastructure to FastAPI, PostgreSQL and Redis caching.
- Python
- FastAPI
- LangChain
- APISIX
- Keycloak
- Redis
- Vector DB
Software Engineer · Adorsho Pranisheba Ltd
Oct 2022 – Aug 2023
Dhaka, Bangladesh
- Reached 95% cattle identification accuracy across 5,000+ records with a CNN-based muzzle-print recognition backend.
- Integrated it into a herd-management system with insurance verification and IoT-based health monitoring.
- Promoted from intern to Software Engineer within 3 months.
- Python
- Django
- OpenCV
- PyTorch
- YOLOv5
- MySQL
- RabbitMQ

Research

Applied ML & computer vision

Alongside engineering, I work on computer-vision and deep-learning research — from biometric recognition to autonomous navigation.

Computer VisionDeep LearningBiometric RecognitionSemantic SegmentationApplied Machine Learning

Publication · IEEE

Development of an Autonomous UAV for Seed and Fertilizer Distribution in Precision Agriculture

Z. A. Jibon, M. A. Adnan, N. T. Nora, M. M. H. Akash, F. Ahammed

Proceedings of the 14th International Conference on Computing Communication and Networking Technologies (ICCCNT), IEEE, IIT Delhi, India, July 2023, pp. 1–5

DOI: 10.1109/ICCCNT56998.2023.10308182

Muzzle-Based Biometric Cattle Identification

Research Engineer · Adorsho Pranisheba Ltd

Oct 2022 – Jul 2023

Proposed an automated cattle-recognition pipeline using FaceNet/ArcFace, reaching 95% accuracy on 5,000+ samples.
Designed YOLOv5 muzzle-region detection and an augmentation pipeline for real-world lighting and angle variability.
Built an end-to-end biometric insurance-verification pipeline that cut manual verification time by 70%.

PyTorch
OpenCV
YOLOv5
FaceNet
ArcFace

Semantic Segmentation for Autonomous Driving in Bangladesh

Undergraduate Thesis · AIUB

Jun 2022 – Oct 2022

Systematically evaluated U-Net, DeepLab and PSPNet on a custom urban-traffic dataset.
Collected and annotated 2,000+ Dhaka traffic images to improve generalization in dense conditions.
Identified key failure modes for urban road scenes and proposed data-centric improvements.

PyTorch
U-Net
DeepLab
PSPNet
Label Studio

Autonomous Mars Rover Navigation System

Software Team Lead · AIUB Robotic Crew

Jul 2021 – Apr 2023

Led the software pipeline for a rover navigation system that placed 11th of 18 finalists (365 teams registered) at IRC 2023.
Integrated RTK-GNSS and AR-Tag localization with ROS-based navigation and sensor fusion.
Developed MobileNetV3 terrain classification and custom real-time inference modules; mentored 4 juniors.

ROS
PyTorch
MobileNetV3
YOLOv5
OpenCV

About

A bit more about me

I'm a software engineer and MSc student in Quantum Information Science at the University of Copenhagen. I like the hard parts of backend work: distributed data, payments that must never be processed twice, and keeping systems observable in production. Most recently I was the primary backend engineer on a payment gateway licensed under Bangladesh Bank rules.

I came up through computer-vision research (cattle biometrics, semantic segmentation, autonomous rover navigation) and still build in that space, which is why my work spans fintech, AI platforms, and IoT. On the side I build open-source systems tools like LiveSQL, Canopy Cloud, and an AI-native VCS in Rust.

Education

M.Sc. in Quantum Information Science

University of Copenhagen, Niels Bohr Institute · 2026–present

B.Sc. in Computer Science & Engineering

American International University-Bangladesh · 2018–2023 · CGPA 3.59/4.0

Achievements

Dr. Anwarul Abedin Leadership Award, AIUB, 2024
11th of 18 finalists (365 teams registered), International Rover Challenge (IRC) 2023, Space Robotics Society
Qualified for the University Rover Challenge (URC) 2022 finals, Mars Society (36 of 99 teams)
IELTS 7.0 / 9.0 — English (Professional)

Tech I work with

Languages

Python
TypeScript
JavaScript
Rust
Go
C#
C++

Backend

FastAPI
Django
Flask
NestJS
.NET
SQLAlchemy
Celery

AI / ML

PyTorch
TensorFlow
OpenCV
LangChain
RAG
Computer Vision
MCP

Data

PostgreSQL
Patroni
MySQL
MongoDB
Redis
Vector DBs

Infra & DevOps

Docker
GitHub Actions
AWS
GCP
NGINX
HAProxy
Keycloak
APISIX
Linux

Observability

Prometheus
Grafana
Loki
Sentry

Frontend

React
Vue
Svelte
Tailwind CSS
Socket.IO