• Computer Science & Engineering graduate, pursuing an MSc in Applied Mathematics & Computational Science at North South University.
• Working at the intersection of Artificial Intelligence, Mathematical Modeling, Systems Architecture & GPU Computing.
• Understanding systems from the ground up — from compilers and numerical ODE/PDE solvers to decoupled distributed web architectures, LLM systems, and CUDA kernels.

• AI & LLM Systems: Retrieval-Augmented Generation (RAG), Fine-Tuning & Quantization, Vector Databases (Qdrant), Agentic AI & Autonomous Tool-Using Systems, NLP Pipelines (LangChain, LangGraph, LlamaIndex).
• System Design & Architecture: Decoupled Headless & Microservice Architectures, RESTful API Design & Contract Specification (DRF, FastAPI), Secure Multi-tenant RBAC, Cryptographic Auth Flows & Attack-Surface Hardening, Relational Data Modeling & Query Profiling (PostgreSQL, MariaDB), Resilient Production Deployment & Asset Pipelines (WSGI / Nginx / Apache).
• Computational Mathematics: Numerical ODE & PDE Boundary Solvers, ADI Schemes, Stiff / Non-Stiff Formulations, Adaptive Solvers (RK45, DOP853, RADAU), Scientific Simulation in MATLAB.
• High-Performance Computing: Dedicated NVIDIA RTX 4070 Ti, CUDA C/C++ & NVCC Kernel Optimization, NVIDIA HPC SDK Accelerated Math, OpenMP Parallelization, Memory Coalescing & Profiling.

• parallel-simulated-annealing-hpc (Ongoing Research · Private)
  High-performance parallel optimization framework investigating distributed and GPU-accelerated Simulated Annealing formulations for large-scale combinatorial spaces. Explores concurrent Markov chain exploration, asynchronous temperature cooling schedules, and shared-memory communication overhead minimization.
  

• steam-trade-card-dropper
  Automated Steam trading card idling utility. Implements Steam client authentication, automated card drop detection, badge tracking, and lightweight background session management.
  

• ARM-ASM-Project
  Domain-specific compiler translating high-level assembly syntax into native ARM-M machine architecture. Implements two-pass assembly parsing, symbol resolution, and binary instruction encoding.
  

• Partial-Differential-Equations
  Computational mathematics and numerical solver suite for boundary-value partial differential equations. Features Alternating Direction Implicit (ADI) methods, parabolic/elliptic solvers, and convergence analysis.
  

• HPC-SDK
  High-performance computing benchmarks and parallel algorithms utilizing the NVIDIA HPC SDK, CUDA C/C++, and OpenMP on the NVIDIA RTX 4070 Ti.
  

• vumeej-jele
  Full-stack advocacy portal and digital archive for the Bhumij Baor Fisherfolk Movement. Architected as a high-performance decoupled system featuring a React 18 + Vite frontend with dynamic multi-template editorial layouts and Bengali typography, backed by a headless Django 5 REST Framework API engineered for resilient production deployment via cPanel Passenger WSGI and MariaDB.
  

Languages

AI & Machine Learning

Systems & GPU Computing

Software & Tools

• System Architecture & Secure Engineering: Architected decoupled, production-grade web systems and headless CMS ecosystems; designed end-to-end security hardening protocols (OWASP compliance, strict role-based access control, cryptographic verification tokens), relational schema optimizations, and custom deployment pipelines across Linux and cPanel/WSGI infrastructures.
• Computational Biology & Genomics: Tracked viral mutations on COVID-19 genome sequencing data via GISAID, blending computational pipelines with biological sequence analysis to extract phylogenetic insights.
• Autonomous UAV Localization & SLAM: Investigated robotic navigation by combining Reinforcement Learning algorithms with SLAM-based spatial mapping for accurate autonomous localization in complex environments.
• Compilers & Architecture: Engineered a custom compiler capable of translating high-level assembly instructions into ARM-M machine architecture, exploring low-level code generation.
• Numerical Solvers & Scientific Computing: Formulated numerical solutions for differential equations: Alternating Direction Implicit (ADI) methods, stiff ODE solvers, and adaptive Runge-Kutta variants (RK45, DOP853, RADAU) in MATLAB.

I write about technology, artificial intelligence, higher education, and economics across prominent national and regional media outlets:

• MSc in Applied Mathematics & Computational Science · North South University (Ongoing)
• BSc in Computer Science & Engineering · North South University
• Competitive Programming: ICPC Regional Contests — Honorable Mentions × 2 (International Collegiate Programming Contest)

_{Still learning. Still building. Still breaking things.}