Bridging Mathematical Theory and Applied Machine Intelligence.
The Research, Analysis & Knowledge (RAK) Lab investigates the algorithmic frontiers of Graph Neural Networks, Multi-Agent Retrieval Systems, and Computer Vision—translating peer-reviewed mathematical models into resilient, local-first production software.
Algorithmic Innovation
Advancing cross-graph attention mechanisms for Graph Similarity Learning, hybrid CNN-ViT ensembles, and multi-modal biological sequence modeling.
Open Reproducibility
Committed to verifiable science. Every published paper and technical note is backed by clean PyTorch architectures, ablation tables, and reproducible pipelines.
Deployable Impact
Shipping models into production ASP.NET Core 8 and FastAPI microservices—powering local Ollama RAG assistants, CrewAI agents, and real-time edge inspection.
Rakibul Islam
AI Software Engineer & Researcher
Rakibul Islam
I work at the intersection of academic machine learning and enterprise software engineering. By day, I build production features for the Alfred and HRM platforms at Genuine Technology Research Limited (GTR) in ASP.NET Core MVC, integrating CrewAI multi-agent workflows, Retrieval-Augmented Generation (RAG), and local Large Language Models.
Through RAK Lab, I conduct independent and collaborative research on how Graph Neural Networks learn structural similarity, how hybrid YOLOv8 and Vision Transformer ensembles localize defects in real time, and how deep networks classify medical radiographs and protein sequences.
Compute & Tooling
Lab Infrastructure
We maintain dedicated local GPU environments and containerized microservice pipelines to train vision and graph networks, run private LLMs via Ollama, and validate reproducible benchmarks.
Local LLM & GPU Rigs
Dedicated CUDA workstations and Ollama inference servers for training GNNs, quantizing YOLOv8 models, and running zero-data-leakage enterprise RAG.
Deep Learning Stack
PyTorch 2.x, PyTorch Geometric, TensorFlow, Hugging Face Transformers, Ultralytics YOLOv8, and OpenCV for vision and graph experimentation.
Agentic & Web Backends
CrewAI, LangChain, LlamaIndex, and ChromaDB orchestrated behind FastAPI microservices and enterprise ASP.NET Core 8 MVC platforms.
Data & MLOps Lineage
Dockerized training environments, Apache Airflow ETL scheduling, PostgreSQL & SQL Server data stores, and reproducible benchmark harnesses.
Explore the Research Ecosystem
Direct DirectoryConference & Journal Papers
Browse 9 peer-reviewed papers and preprints with BibTeX citations.
02 // Sponsored R&DProposals & Grants
Active multi-year grants, industrial partnerships, and archived awards.
03 // SupervisionOpen Thesis Topics
Curated BSc, MSc, and PhD research problems in GNNs, Vision, and RAG.
04 // Knowledge BaseTechnical Notes & Essays
Mathematical derivations, PyTorch implementations, and MLOps guides.