Ingénieur IA appliquée | Expert en Callbots et Voicebots

Welcome to my GitHub! I design and develop intelligent voicebots and callbots using advanced NLP techniques and LLMs, integrating telephony systems with automated workflows.

• Independent AI Engineer specializing in callbots, voicebots, and conversational agents.
• Experienced in deploying and scaling telephony infrastructure (LiveKit, SIP, Telnyx, Twilio) on cloud platforms (AWS).
• Skilled in Natural Language Processing (NLP): Sentiment Analysis, PoS Tagging, NER, Question Answering, Summarization, and ASR.
• Passionate about building automated and intelligent systems that optimize business processes and reduce operational costs.

• Callbots / Telephony: LiveKit, Telnyx, Twilio, VAPI, pipecat, n8n, LangGraph, SIP Trunking
• Backend & DB: Flask, FastAPI, PostgreSQL, ChromaDB
• NLP & LLMs: Transformers (BERT, GPT), LoRA, QLoRA, PEFT, RAG
• Data & Visualization: Pandas, NumPy, Matplotlib
• Evaluation Metrics: F1, WER, BLEU, Precision, Recall
• Languages: French (C1), English (C1), Arabic (C2)

• Semi-supervised Word Sense Disambiguation (K-means). Repo: https://github.com/Caegi/Semi_Supervised_Word_Sence_Disambiguation
• Retrieval-Augmented Generation (RAG) pipelines
• Fine-tuning Transformers with LoRA / QLoRA

Developed several AI-powered callbots using LiveKit, Telnyx, and n8n, integrated with Google or Outlook Calendars and deployed on AWS or LiveKit Cloud. Projects include:

• BarberShop Callbot – automated phone-based appointment scheduling and synchronization with Google Agenda.
• Diamanda – internal and external call management system with intelligent routing and Outlook integration. (W3TEL)
• Box Self-Storage Callbot – outbound system for client surveys and notifications.
• Company Data Collection Callbot – inbound assistant for collecting company IT data for resale estimation.

Repo: https://github.com/aspirant2018/niqatech-backend Development of an innovative application enabling teachers to input students’ grades using their voice. The system uses speech-to-text (STT) for audio transcription, then leverages LangChain to automatically extract structured information (student, subject, grade) and insert it into a grade file.

• Technologies: Python, FastAPI, LangChain, Whisper (STT), PostgreSQL, Excel
• Extraction of structured information from transcriptions using LangChain (structured output)
• Integration of a LangChain-based chatbot component with embeddings, chunking, vector indexing (Qdrant), and re-ranking for accurate responses
• Implementation of interactive API documentation with Swagger (OpenAPI)
• Development of a simple and intuitive React.js interface to streamline grade entry

• Email: mazouzceminfo@gmail.com
• LinkedIn: linkedin.com/in/abderahim-mazouz-6952b6141
• GitHub: github.com/aspirant2018

"Automating human interaction intelligently, one call at a time."