Md Hasibul Amin
Software Engineer — Distributed Systems, Cloud Infrastructure & AI/ML
Houston, TX, USA
Email ·
Google Scholar ·
GitHub ·
LinkedIn ·
Hugging Face
About
I’m a software engineer with 10+ years of experience designing and building distributed backend systems, cloud infrastructure, data platforms, and production-critical services.
My engineering work spans event-driven architectures, Kafka and CDC pipelines, CQRS, scalable APIs, multi-tenant systems, Kubernetes, infrastructure-as-code, observability, incident response, and production reliability across AWS and Azure.
I have also led and mentored engineering teams, owned backend and database architecture, and worked on systems serving hundreds of thousands of daily requests.
Alongside my full-time engineering work, I conduct and collaborate on applied AI/ML research in continual learning, model reliability, NLP, medical imaging, and LLM adaptation and evaluation.
I earned an M.S. in Computer Science from the University of Houston, where I worked on GPU-accelerated medical-imaging and applied AI systems.
Engineering Experience
Software Engineer
Crane Worldwide Logistics · Houston, TX · Jul 2023 — Present
- Designed and built from scratch a resilient Kafka-based billing platform automating invoice validation, document retrieval and assembly, downstream delivery, and notifications
- Implemented idempotent processing, retry policies, and DLQ-backed recovery, with 99% of requests completing within 90 seconds
- Built Kafka CDC pipelines synchronizing tens of millions of production records and propagating ongoing changes within a 30-second latency SLA
- Designed CQRS-based PostgreSQL read models to offload read-heavy workloads from Oracle, improving query response time by approximately 15–18%
- Serve as production DRI across mission-critical platforms, handling incident response, debugging, root-cause analysis, and reliability improvements
- Lead architecture, design, and code reviews for C#/ASP.NET and F# services across AWS and Azure
- Build secure multi-tenant APIs with configuration-driven behavior, database-per-client isolation, and least-privilege RBAC
- Work with CI/CD, Terragrunt, Bicep, Kubernetes, OpenTelemetry, and distributed tracing
- Use Claude Code, GitHub Copilot, and MCP-based workflows for requirements analysis, implementation, testing, debugging, and code review
Lead Software Engineer
Truck Lagbe Ltd. · Dhaka, Bangladesh · Sep 2019 — Jul 2021
- Led and mentored a 12-engineer cross-functional team while owning backend and database architecture for a freight marketplace handling 400K+ daily requests and 5K+ concurrent users
- Built a multilingual geo-search engine mapping informal landmarks to formal addresses using trie-based search, inverted indexing, spatial hashing, and fuzzy matching
- Achieved approximately 95–97% search accuracy and reduced external mapping costs by approximately $35K annually
- Designed high-availability database architecture using CQRS, replication, age-based trip partitioning, caching, and recovery workflows
- Architected the core backend and reusable Node.js framework used across engineering teams
- Built an adaptive pricing engine using distance, time, demand, weather, vehicle type, and historical trip data; rollout testing increased successful bookings by 20–23%
- Migrated production workloads to AWS and Kubernetes, reducing deployment time from 2–3 hours to approximately 20–30 minutes
- Served as Acting CTO for two months, setting engineering priorities and leading a major cross-platform release
Associate Senior Software Engineer
CSL Software Resources Ltd. · Dhaka, Bangladesh · Feb 2017 — Aug 2019
- Built a reusable C++ access-control library for C#/ASP.NET applications, reducing development and maintenance effort by approximately 10%
- Developed ERP and POS systems used by 400+ companies and 250+ retail outlets
- Optimized APIs and SQL workloads, improving dashboard response time by approximately 35%
Software Engineer
GOBD (Startup) · Dhaka, Bangladesh · Jan 2015 — Jan 2016
- Built route-optimization logic combining live tracking, traffic-signal, mobile-app, and third-party location data
- Weighted data sources by priority to identify alternate routes under changing traffic conditions
Research Experience
Independent AI/ML Research Collaborator
Late 2024 — Present
Conducted alongside full-time software engineering work with multi-institution collaborators.
- Develop and evaluate continual-learning methods in Python and PyTorch to reduce catastrophic forgetting
- Study model-update reliability and failure modes in replay-based continual learning
- Identified a readout-dependent failure mode that can reverse the measured benefit of knowledge distillation under controlled multi-seed evaluation
- Collaborate on research in breast-cancer segmentation, medical LLM fine-tuning, NLP, and trustworthy AI
- Built a BERT-based pipeline for COVID-19 vaccine conspiracy and misinformation detection using Google Perspective API signals for content-quality analysis
Graduate Research Assistant
University of Houston — Imaging Innovations Laboratory · Houston, TX · 2021 — 2023
- Built an automated GPU-accelerated CT acquisition and reconstruction pipeline in Python
- Integrated 360° acquisition and precise positional control
- Reduced end-to-end processing time by approximately 35–40%
- Worked on biomedical image analysis and applied AI research
Selected Engineering & AI Projects
Designed production-critical event-driven workflows using Kafka, CDC, C#, F#, SQL, PostgreSQL, Oracle, AWS, Azure, and Kubernetes.
Focus areas:
- Event-driven architecture
- Idempotency
- Retry and DLQ recovery
- Data synchronization
- Distributed tracing
- Production reliability
- Multi-tenant service design
Proprietary implementation. Public descriptions intentionally exclude confidential business details.
Continual Learning and Model Reliability
Research repository:
Continual-Learning-UsingInterClusterDistance
Preprint:
Failure Modes of Always-On Inter-Cluster Repulsion in Replay-Based Continual Learning
Research areas include:
- Catastrophic forgetting
- Replay-based continual learning
- Knowledge distillation
- Geometric representation analysis
- Multi-seed evaluation
- Ablation studies
- Reproducible experimentation
- Failure-mode analysis
First-author research on detecting COVID-19 vaccine conspiracy and misinformation content.
- BERT-based NLP experimentation
- Manually labeled social-media dataset
- Google Perspective API signals
- Content-quality and misinformation analysis
- Public dataset release through Hugging Face
Hugging Face Dataset ·
arXiv Paper
Technical Expertise
Languages:
Python, C#, F#, C++, Java, JavaScript, SQL
Backend & Distributed Systems:
Kafka, CDC, Microservices, CQRS, REST APIs, Event-Driven Architecture, Workflow Orchestration, Multi-Tenant Systems, Idempotency, Retry/DLQ, High Availability, System Design
Cloud & Platform:
AWS, Azure, Kubernetes, Docker, Terraform/Terragrunt, Bicep, Lambda, Azure Functions, CI/CD
Data & Observability:
PostgreSQL, Oracle, SQL Server, MongoDB, Redis, OpenTelemetry, Distributed Tracing, Structured Logging, Incident Response, RCA
AI/ML & Search:
PyTorch, TensorFlow, Continual Learning, NLP, BERT, LLMs, RAG, Model Evaluation, Information Retrieval, Trie, Inverted Index, Spatial Hashing, Fuzzy Matching, OpenCV
AI-Assisted Development:
Claude Code, GitHub Copilot, MCP workflows
Selected Publications
2027 — Under Review
Residual Stream Rebalancing: Training-Free Hallucination Mitigation in Vision-Language Models
Niamul Hassan Samin, Md Hasibul Amin, et al.
WACV 2027 — Under Review
OpenReview
When Detector-Based Grounding Metrics Measure Vocabulary: A Cautionary Audit of Entity Claims in Video-QA Reasoning Traces
Md Ashikur Rahman, Md Hasibul Amin, et al.
WACV 2027 — Under Review
OpenReview
2026
Boundary-Aware Deep Learning Framework for Breast Cancer Image Segmentation: Enhancing Tumor Delineation Through BIOU-Driven Evaluation
IEEE CICN 2026
IEEE Xplore
Evaluating Parameter-Efficient Fine-Tuning of Open-Weight LLMs for Medical Question Answering
IEEE CICN 2026
IEEE Xplore
Multilingual Financial Fraud Detection Using Machine Learning and Transformer Models: A Bangla-English Study
IEEE CICN 2026
IEEE Xplore
Preprints
Failure Modes of Always-On Inter-Cluster Repulsion in Replay-Based Continual Learning
Md Hasibul Amin, Tamzid Tanvi Alam
arXiv ·
GitHub
Detecting Conspiracy Theory Against COVID-19 Vaccines
Md Hasibul Amin, et al.
arXiv ·
Dataset
Professional Service
- Technical Program Committee Member, IEEE CICN 2026
- Peer Reviewer, PLOS ONE
Education
Master of Science in Computer Science
University of Houston · 2023
Relevant areas: Machine Learning, Deep Learning, NLP, Cloud Computing, Digital Image Processing
Bachelor of Science in Computer Science and Engineering
American International University-Bangladesh (AIUB) · 2015
Relevant areas: Algorithms, Data Structures, OOP, C++, C#, Java
Email: hasibulamin.cse@gmail.com
Location: Houston, TX, USA
Open to opportunities and collaborations in distributed systems, cloud infrastructure, AI infrastructure, continual learning, model reliability, and applied AI research.
Last updated: October 2026