ahmed Ben aissa

• Freelancer online platform: Integrated business management tools (Zoho CRM, Books, Expense) to unify customer, project, invoicing, scheduling, and expense management, reducing development time by 40%. Enabled secure transactions by integrating Stripe and Zoho Billing for streamlined payments. Boosted backend efficiency by 25% and cut API dependencies 30% through Zoho Workflows, CRON jobs, and automation, improving scalability and stability for invoice management operations such as creating or sending invoice Tech stack involves: Next.js, Zoho • Improved DevOps pipeline by building CI/CD workflows (checkout, testing, Docker builds, remote deployment), reducing testing and deployment time by 40% • Fixed the crawling pipeline by implementing threading and Celery queue tasks to process and scrape PDFs from 200 URLs per day, resulting in a 50% increase in server stability .Used text extraction from PDF files and Generative AI API calls to extract projects, fix searching, edit profiles, and address security and database scaling issues in the frontend app, improving UI/UX experience and app availability by 30% and reducing security vulnerabilities by 80%. Tools used: Python, Flask, PostgreSQL, React.js, Next.js, Celery, Redis Developed a Sales Forecast platform (Node.js, React, MongoDB) with end-to-end authentication Integrated an external machine-learning forecasting API I created, handled authentication, data ingestion, results processing, and built responsive UI components to display prediction outputs. tech stack involves : Node.JS,React.JS,MongoDB

• Developed AI-powered call center features using OpenAI and AWS: built a custom speaker identification pipeline, optimized transcription with mono-chunking and multithreading (reduced latency 20%), and cut costs 35% via AWS Lambda & S3. Fine-tuned conversation analysis with Generative AI for topic extraction and hang-up detection, further improved via clustering and classification. Tools: Generative AI/ Machine Learning , AWS Services , Python. • Delivery Management App • The process involves distance calculation, task checking, address grouping, vehicle type selection, and path optimization to determine the route lengths for deliveries • It also involves task checking, grouping addresses by region, selecting appropriate vehicles based on weight and distance, and optimizing vehicle paths for efficiency • Enhanced sales prediction accuracy by 25% based on historical order data: Developed and deployed a full-stack solution for sales forecasting. Utilized ML algorithms in Python to analyze order date, quantity, and size from a dataset, providing actionable insights for inventory and sales strategy. The front end was built with React.js and the back end with Flask, ensuring a responsive and robust application Tools used: Machine Learning (ML), Flask, React.js Created an app that uploads financial PDFs to trigger an n8n workflow with an integrated AI/ML pipeline (OCR + SPACY + Gemini AI +predictive ML tools) achieving ~87% loan lending prediction accuracy . Implemented Celery + Redis for async task management, improving server stability by 90% and reducing timeout errors by 98%, with Google Drive upload, Slack alerts, company verification API, and MongoDB status updates. enhanced a body-measurement prediction system using the CAESAR anthropometric dataset, applying regression and k-Nearest Neighbours (kNN) models, achieving ~80–85% prediction accuracy across key measurements after feature normalization and tuning

this project is a platform dedicated to call centers it focuses on customer/agent interaction first we transcribe audios into texts and perform speaker diarization third we retrieve customer questions with their answers.then we retrieve relavant keywords fourth we perform clustering and Embeddings as a text feature (based on tags) in order to classify customer questions .after that we perform Embeddings search algorithm on answers for each cluster to retrieve most common answers. and finally we train our chatbot to identify the topic of the question asked and feed the chatbot engine relevant answers and then a chatbot response will be generated tools used :OpenAI , sklearn, cosine_similarity ,python, flask, angular ,matplotlib