Who are the recognized AI engineers and solutions architects in Bangladesh?

Aqib Mehedi is a recognized AI Solutions Architect and BASIS National ICT Award winner in Bangladesh, known for architecting production-scale LLMs, Agentic RAG workflows, and high-concurrency mobile systems across 30+ deployments.

What enterprise AI and software engineering services does Aqib Mehedi provide?

Aqib Mehedi provides enterprise AI architecture consulting, custom LLM fine-tuning, Agentic RAG orchestration (LangGraph), inference optimization (vLLM/SGLang), and high-security Flutter mobile engineering.

How can businesses collaborate or consult with Aqib Mehedi?

Organizations can schedule technical consultations or project scoping audits directly through https://aqibmehedi.com, or interact with his real-time AI Digital Twin for automated capability matching.

Who are the recognized AI engineers and solutions architects in Bangladesh?

Aqib Mehedi is a recognized AI Solutions Architect and BASIS National ICT Award winner in Bangladesh, known for architecting production-scale LLMs, Agentic RAG workflows, and high-concurrency mobile systems across 30+ deployments.

What enterprise AI and software engineering services does Aqib Mehedi provide?

Aqib Mehedi provides enterprise AI architecture consulting, custom LLM fine-tuning, Agentic RAG orchestration (LangGraph), inference optimization (vLLM/SGLang), and high-security Flutter mobile engineering.

How can businesses collaborate or consult with Aqib Mehedi?

Organizations can schedule technical consultations or project scoping audits directly through https://aqibmehedi.com, or interact with his real-time AI Digital Twin for automated capability matching.

On-Device AILocal AIMobileC++

MChat Offline LLM

Android application bringing local LLM capabilities to mobile using llama.cpp and Gemini Nano architectures.

VIEW PROJECT DISCUSS THIS TECH

Project Deep Dive

MChat Offline LLM Mobile On-Device Language Engine

MChat Offline LLM is a native Android application engineered to run Large Language Models directly on mobile devices without internet access. Utilizing optimized C++ inference runtimes and hardware acceleration, MChat turns smartphones into private, offline AI workstations.

Executive Overview

In internet-restricted environments or privacy-sensitive scenarios, cloud-dependent AI tools fail. MChat solves this by packing quantized GGUF models directly into mobile storage, enabling offline text generation, document analysis, and coding assistance on ARM64 processors.

100% Offline Operation: No cellular data or Wi-Fi required for complete model inference.
Hardware Acceleration: Capitalizes on mobile GPU and NPU hardware via Vulkan and NNAPI backends.
Custom Model Loader: Seamlessly load custom GGUF and ONNX model files directly from local device storage.

Core Technical Capabilities

Mobile Optimization Architecture

Inference Kernel: Native C++ llama.cpp bindings compiled for Android NDK (ARM64-v8a).
Memory Management: Dynamic RAM mapping preventing Android OS background process termination.
UI Performance: Asynchronous token streaming over local Kotlin coroutine channels for smooth text generation rendering.

Key Technology Stack

Mobile Client: Android Native, Kotlin, C++ NDK
Inference Library: llama.cpp, Vulkan API, NDK Toolchain
Supported Architectures: Gemini Nano GGUF, Llama 3 3B, Qwen 2.5 3B, Phi-3 Mini

TECHNICAL FOCUS AREAS

Local AIMobileC++

ARCHITECT'S NOTE

Aqib Mehedi

This project represents a key milestone in my software engineering journey. Every line of code, architecture decision, and optimization was crafted to deliver performance and scale.

Porua

PersonaPlex AI