As part of the Support team, we focus on automating and enhancing customer experiences through cutting-edge AI solutions. I have successfully trained and deployed a Text-to-Speech (TTS) model, now integrated into the Driver's Navigation Application.I am also leading the development of an AI-driven educational call system that communicates directly with drivers. This system leverages Speech-to-Text, TTS, and Large Language Models (LLMs), combined with advanced prompt engineering, to explain company plans and respond to driver inquiries in realtime.My work enables scalable, automated support solutions that improve operational efficiency and elevate the user experience.

Time Series ForecastingTime Series Analysis of stock’s trading Volume in the Tehran Stock Exchange.Forecasting trading Volume for each stock in Daily and 10 Minutes periods with LSTM and Auto ARIMA models.Improving the Mean Absolute Percentage Error (MAPE) of Daily Volume prediction by 5%.Deployment of online models on the production server using RESTful API, micro-services, RabbitMQ, and MongoDB.Face Liveness Detection, Face Detection and RecognitionResearch & Development of "Electronic Know Your Customer" (EKYC) service based on Biometrics.Training “Multiresolution Knowledge Distillation for Anomaly Detection” and “Single-Side Domain Generalization” for Face Liveness Detection using PyTorch.Improving TPR@FPR=%5 of the Liveness Detection model from 55% to 94% on the custom dataset.Identity verification with Face Detection and Verification.Research and development of Challenge based Liveness Detection with Head Pose Estimation.Research and development of Challenge based Liveness detection with Speech to Text service.Deployment of models with RESTful API and micro-services.

Research & Development of Video analytics and Facial Recognition systemsTraining “YOLOv3” Object Detection Model for custom Dataset.Improving the “mAP” of the trained model on custom Dataset from 26.2% to 50.1% while maintaining this measure on COCO Dataset on 54%.Implementation of multi-class object tracker in Object-Oriented Python using PyTorch.Training Gender Classification System based on Face image using handcrafted features including Histogram of Gradients (HOG), CrossMax, etc., and Support Vector Machine.Implementation of deep learning models and algorithms in C++ using OpenCV and TensorFlow for productions.Camera Calibration to optimize Face Detection and Recognition system based on Geometrical and Lighting situations.

Electronic designer and embedded system programmer

A deep study of “Hands-On Machine Learning with Scikit-Learn and TensorFlow”.Teaching an online course of Machine Learning including +60 hours of fundamental concepts & implementation of algorithms with Python & Scikit-learn.

Electronic & Robotic Teaching, Designing Electronic Boards, Programming Embedded Systems.