Bechir B.

AI/ML Engineer specializing in Data Science, Machine Learning, Deep Learning, Data Analytics. 🌍 Available across US, Canada, Europe, Australia, Middle East 📚 Publications: @Bechir Ben Tekfa, et al., DEXA 2025: Advanced Defect Detection in Industrial Imaging @Bechir Ben Tekfa, et al., KES 2025: Hybrid AI Systems for Renewable Energy Inspection ⏱️ 300+ billed hours delivering multi-domain AI & ML systems I design, implement, and deploy intelligent systems across multiple domains: computer vision, LLMs, time-series, audio, graph-based reasoning, and AI-driven automation. My focus is on real-world constraints, building robust, scalable, multi-modal systems that solve complex problems at once. 🧪 Core Expertise & Research Domains 1️⃣ Computer Vision & Visual Intelligence • Defect detection, anomaly localization, and automated inspection • Multi-modal inputs (EL, IR, RGB) with noise, bias, and real-world variability • Object detection, segmentation, hybrid CNN-attention architectures • Weakly supervised, long-tail, and low-data regimes • Multi-scale feature fusion, interpretability, and error analysis • Deployment-aware models for real-time, edge, and constrained devices 2️⃣ Machine Learning Foundations • Supervised, semi-supervised, self-supervised, representation learning • Time-series modeling, anomaly detection, forecasting • Graph-based learning, relational reasoning, and knowledge graphs • Mathematical rigor: optimization, regularization, uncertainty, generalization • Classical ML + deep learning hybrids for structured and unstructured data 3️⃣ LLMs & Reasoning Systems • Multi-step reasoning and orchestration pipelines • Retrieval-Augmented Generation (RAG) for domain-specific knowledge • AI agents for decision-making, planning, and automation • Reliability, hallucination control, evaluation, and system boundaries 4️⃣ Audio & Signal Processing • Speech-to-Text (STT) and Text-to-Speech (TTS) integration • Spectral, temporal, and multi-modal embeddings • Cross-modal pipelines (audio + vision + text) 5️⃣ AI-Driven Automation & Orchestration • Web scraping and data collection (Selenium, Scrapy, BeautifulSoup) • Workflow automation and orchestration (n8n, Twilio, Zapier-like pipelines) • Automated labeling, ETL workflows, ML pipeline orchestration • Continuous data ingestion and pre-processing for multi-domain AI systems • CI/CD, experiment tracking, reproducibility, and monitoring 6️⃣ Edge AI & Deployment Systems • NVIDIA Jetson and resource-constrained inference • Model optimization: pruning, quantization, TensorRT, ONNX • Dockerized deployments for reproducibility and scalability • Latency, memory, and energy trade-off analysis 🧰 Technical Stack AI / ML: Python, PyTorch, TensorFlow, scikit-learn, Hugging Face, OpenAI API Computer Vision & Signals: CNNs, Detection, Segmentation, Spectral Features LLMs & Automation: RAG, Agents, Orchestration, Evaluation Pipelines Data & Systems: Pandas, NumPy, SQL, PySpark, ETL pipelines Automation & Integration: Selenium, Scrapy, BeautifulSoup, n8n, Twilio Deployment & MLOps: Docker, ONNX, TensorRT, Jetson, CI/CD pipelines