Building the future of Domain-Specific LLMs
Specialized in High-Performance Inference & Distributed Training Pipelines
|
Full-Stack AI & MLOps Engineer dedicated to building scalable, high-performance AI systems. Bridging the gap between Complex Deep Learning and Production-grade Infrastructure. |
-
🧠 Large Language Models (LLMs)
- Agentic Systems: Crafting Multi-Agent Architectures & Tool Usage.
- Alignment: Expert in Instruction Fine-Tuning and Alignment (DPO, PPO, RLHF).
- Pipelines: Building complete pipelines from Pretraining to Performance Monitoring.
-
⚡ Model Optimization & Efficiency
- Efficient Finetuning: Specialized in LoRA, QLoRA, and Adapter Fusion.
- Distributed Training: Scaling via DDP, Tensor Parallel, and FSDP.
- Quantization: Reducing latency using AWQ, GPTQ, and ONNX.
-
🏗️ MLOps & Production Serving
- Serving: High-throughput deployment using vLLM, NVIDIA Triton, and KServe.
- Infrastructure: Kubernetes, Docker, Knative, and Fabric Manager.
- APIs: Developing robust backends with FastAPI and Flask.
-
👁️ Computer Vision Intelligence
- Advanced Segmentation, Object Detection, and OCR for multimodal integration.
| Event | Rank | Description | Links |
|---|---|---|---|
| Liltbench (Terminal Bench) | 🥇 #6 | Secured 6th place globally solving complex AI agent terminal tasks and command-line automation benchmarks. | Leaderboard |
"I don't just train models; I design AI ecosystems that think, reason, and scale. Transforming GPU raw power into seamless, real-world value."
| 🌟 Project | 🧩 Description | 🔗 GitHub |
|---|---|---|
| Recommandation System | LLM + RAG + CV pipeline | Link |
| AI Image Restoration App | Denoising & Super-Resolution | Link |
| OCR Scanner App | Multi-page scan to PDF with Tesseract | Link |
| Medical Imaging Diagnosis Agent | Streamlit + LLM + CV | Link |
| AI News Agent Team | Multi-agent educational content generator | Link |
| Medical Image Chatbot | RAG+Chorme db | Link |
| Smart-Hotel-Reservation-System-AI-Powered | LightGBM OCR Agent | Link |
| Attentition Library for Computer vision | Several Attentition Techniques in Computer Vision | Link |
| CAMELYON17-Breast-Cancer-Detection | diagnose breast cancer | Link |
| CGIAR Root Volume Estimation Challenge | challenge to predict root volume from plant images using multi-modal data | Zindi Link |
| ZINDI-Mapping-Floods-in-South-Africa | developing machine learning models to detect and map flood-affected | Link |
| Text-to-MySQL Query Generator (LLM-Powered) | converts plain English (or Arabic) text instructions into executable MySQL | Link |
| AI Bubble Sheet) | intelligent Bubble sheet app using OpenCV, LSD edge detection,Flutter. | Link |
| RLHF_notebooks | implementation for Reinforcement Learning from Human Feedback (RLHF) | Link |
| Image Editing Pro | Image Editing Professional using Pytohn and Gradio | Link |
| image denoise pro | Image Denoising Using MiXnet neural network | Link |
| Image Super resolution | Using GPEN for face enhancement and via FastAPI | Link |
| Smart Image Sharpnees | modify huging face space to work through Fast API | Link |
| Multimodal Remote Sensing Analysis(GeoCLIP & Flamingo) | Multimodal Learning for Satellite Imagery and Geospatial Intelligence | Link |
| AI_Recommandiation_System | AI Recomandation System for Sales | Link |
| Defocus Debluring | Remove deblueing from image using Restomer Net | Link |
| person_detection_from_video | Detecting Person from Camera in Secueity purpose | Link |
| Disease-Classification Kaggle | Plantes Disease-Classification | Link |
| Smart Manufacturing Machines Efficiency Prediction | predicting the operational efficiency of industrial manufacturing machine | Link |
| AI-Music-Composer | AI Muisc Composer from End to end K8,Jenkins,GCP | Link |
| msda | real-time anomaly detection from high-dimensional heterogeneous/homogeneous time series | Link |
| Time Series Forecasting with Lag Llama | Time Series Forecasting with Lag Llama | Link |
| luminol | time series data analysis. | Link |
| Rain Predication | Australia-Weather-Rain-Prediction | Link |
| Well-Performance-Prediction-and-Anomaly-Detection | predicting the performance of wells in terms of gas, oil, and water production | Link |
| AI-Powered Assessment of Vegetable Cutting Techniques Using Multi-Modal Computer Vision Models | Assessment of Vegetable Cutting Techniques | Link |
| Fire Detection System | real-time fire detection in images and videos using a fine-tuned YOLO model | Link |
| Tennis-Analysis-Application | tennis match analysis system | Link |
Python • Modern C++ • SQL • Bash / Shell
- Models:
PyTorch•TensorFlow•HuggingFace•Transformers - Agents:
LangGraph•CrewAI•AutoGen•Agno•LangChain - Techniques:
RAG•CAG•DSPy•Prompt Engineering•Agentic Workflows
- Alignment:
SFT•RLHF•DPO•PPO - Optimization:
Quantization (AWQ/GPTQ)•LoRA/QLoRA•ONNX•Model Distillation
- Serving:
vLLM•NVIDIA Triton•KServe•Knative - Orchestration:
Kubernetes•Docker•ArgoCD•FluxCD - Pipelines:
MLflow•Flyte•Airflow•DVC•Jenkins - Cloud:
GCP (Google Cloud)•AWS•Fabric Manager
- Vector DBs:
Pinecone•Milvus•ChromaDB - Databases:
PostgreSQL•MongoDB•MySQL•Redis - Big Data:
Apache Spark•Kafka
OpenCV • Image Processing • Object Detection • Segmentation • OCR
Git • Linux • Flutter • FastAPI • Flask • Streamlit
- Frameworks:
LangSmith•DeepEval•OpenAI Evals - Methods:
LLM-as-a-Judge•Pairwise Evaluation•Human Feedback - Metrics:
Hallucination Detection•Faithfulness•Groundedness - Benchmarks:
Terminal Bench•Harbor•Agent Evaluation
- GPU:
CUDA•cuDNN•NCCL - Inference:
TensorRT•TensorRT-LLM - Profiling:
Nsight Systems•Nsight Compute - Optimization:
Kernel Optimization•Mixed Precision•Quantization
- Training:
DDP•FSDP - Inference:
Tensor Parallelism•Pipeline Parallelism - Communication:
NCCL•NVLink
- Retrieval:
Hybrid Search - Ranking:
Reranking - **Embedding Models`
- **Semantic Search`
Certified by Micro1 after successfully passing their AI technical evaluation process.
- Outstanding performance in Micro1 AI Interview
- Added to Micro1's vetted talent pool
- Evaluated on AI engineering, problem-solving, and technical communication
| 🧠 Title | 🔗 DOI / Link |
|---|---|
| Paddy Leaf Disease Classification using Enhanced CoAtNet + MobileNetV2 | 🔗 Zenodo |
| AI-Powered Assessment of Vegetable Cutting Techniques | 🔗 DOI |
| YOLO + Kalman Filter for Tennis Ball Tracking | 🔗 DOI |
📧 Email: ML_mohamed_ahmed_87@outlook.com
🌍 LinkedIn: mohamed-ahmed-700019a5
🧠 Blog: DeepLearning4ComputerVision
🧾 Orcid: 0009-0001-3342-7233
💼 Turing: Profile
“Turning ideas into AI-driven realities — one model at a time.” 🚀