Welcome to my Digital Space

AHMED MOHAMED

AI Engineer & RAG Systems Architect

Undergraduate AI Engineering researcher at Bahçeşehir University. Specializing in PyTorch deep learning systems, real-time computer vision structures, and deploying resilient enterprise-grade multi-agent RAG ecosystems.

View Projects Collaborate
Gemini Core Active
System Creed

Core Engineering Philosophy

Mathematical Rigor

I believe in deep structural analysis over blind parameter tuning. Every layer, loss coefficient, and vector dimension in my systems is mathematically calculated to maximize data throughput and optimization convergence.

Bare-Metal Devotion

Efficiency starts at the hardware layer. Whether it is compiling custom CUDA kernels, optimizing deep tensor layouts, or running hyper-lean workloads on native Arch Linux systems, I build to eliminate every millisecond of container latency.

Agentic Coexistence

Simple API wrappers are obsolete. The future belongs to complex, autonomous multi-agent networks that dynamically route queries, use specialized CLI tools, index vector weights, and self-correct their logic loops in real-time.

Journey

The Engineering Timeline

[2023 - Present]

Bahçeşehir University (BAU)

AI Engineering Undergraduate

Focusing on deep neural networks, computational statistics, computer vision, and multi-agent RAG vector databases. Actively researching custom loss functions for spatio-temporal layers.

[PyTorch CV Research]

Spatio-Temporal Sentence Recognition

Deep Computer Vision Engineering

Architected a deep learning network utilizing PyTorch that intercepts high-definition OpenCV silent video streams and translates sequence lip-gestures into grammatical sentences using 3D CNNs + Bidirectional GRU layers.

[Enterprise AI Deployment]

Production Agentic Orchestration

FastAPI Microservices Systems

Engineered a production-ready asynchronous WhatsApp platform managing concurrent Webhook messages, powered by a FastAPI backend, Supabase pgvector embedding indexes, and automated n8n reasoning loops.

Engineering

Featured AI Projects & Technical Specs

WhatsApp AI RAG Platform

Arch: Asynchronous FastAPI + Supabase pgvector + n8n webhook routing.

A production-grade, highly concurrent conversational assistant integrated into WhatsApp. Implements cosine similarity search over vector dimensions, handles automatic chat session caching, and executes workflow tool-calling dynamically.

FastAPI React Supabase n8n RAG Session Caching

Lip Reading Recognition System

Model: 3D ResNet + Bi-GRU + CTC Loss Optimizer (PyTorch).

An advanced spatio-temporal computer vision system that interprets spoken words from silent video frames. Preprocesses frames at 25fps via OpenCV, feeds normalized tensors into PyTorch, and maps sequence alignments with Connectionist Temporal Classification (CTC) loss, achieving a high 92.4% validation accuracy.

PyTorch OpenCV 3D ResNet Bi-GRU Sequence CTC Loss

Intelligent Academic Assistant

RAG: LlamaIndex + LangChain + Parent-Child Node Chunk Indexing.

An agentic study buddy executing cognitive retrieval over academic textbook corpuses. Features hybrid semantic-lexical search indices, dynamic metadata context enrichment (syllabus alignment), and multi-step reasoning chains for complex queries.

LlamaIndex LangChain Vector Embeddings Node Indexing Agent Tools
Skills

Technical Stack Registry

Languages

Python (Deep Stack)
SQL (PostgreSQL)
C# (.NET Core)
C++ (Algorithms)
Go (Concurrency)

AI & Backends

PyTorch (CNN/RNN/GRU)
FastAPI (ASGI Microservices)
ASP.NET Web API
React (SPAs)
LangChain
LlamaIndex (Vector Indexes)

Systems & DevOps

Linux (Arch Core)
Docker Containerization
WSL (Windows CLI)
Supabase DB
Collaboration

Secure Connection Portal