Hossein
Jandaghi

Hi! I am a specialized AI & Machine Learning Engineer with over 4 years of hands-on experience building, optimizing, and deploying intelligent systems. Unlike many developers who strictly work in academic or sandbox environments, my primary focus is on production-ready AI. The machine learning models and data pipelines I have developed are currently live and actively serving enterprise applications with over 5 million active users. My Core Areas of Expertise: Natural Language Processing (NLP) & Audio: Building highly accurate, offline Speech-to-Text (STT) engines and text-analysis pipelines (sentiment analysis, behavioral scoring, and automated summarization). Predictive Analytics & Financial Forecasting: Developing robust time-series forecasting models using deep learning and macroeconomic feature engineering, integrated with real-time news crawlers for market sentiment analysis. Data Science & Customer Intelligence: Designing intelligent customer segmentation systems (

https://www.linkedin.com/in/hossein-jandaghi/

https://github.com/Jandaghi

Work experience

saina pardazesh gostaresh shargh

February 2023 – Current time

Job title

AI Engineer

Work experience & achievements

Environment: High-traffic Financial & Utility Applications (5M+ Active Installs) Work Description: Led the end-to-end research, development, and deployment of applied artificial intelligence solutions. My primary responsibility was to design scalable machine learning models that solve complex business logic, process massive datasets, and directly improve product efficiency and user experience. Key Achievements & Delivered Projects: Offline NLP & Speech Recognition (iCRM): Engineered a fully offline, highly optimized Speech-to-Text (STT) engine achieving an exceptional 3.5% Word Error Rate (WER). Designed a custom NLP pipeline that automatically processes thousands of support calls to extract call summaries, categorize topics, and dynamically score customer service agents on soft skills (politeness, empathy, active listening) on a 0-100 scale. Intelligent Customer Segmentation (iRFM): Transformed traditional marketing analytics by developi

Education

NIT University

Graduation year: 2020

Level of study

PGCE

Major / Field of study

AI Engineer

Hossein Jandaghi

(0)

I will develop Custom Machine Learning, NLP, and Data Science Models

Web, Mobile & Software Dev

Python

Back End Developer

$25

$25/hr

More info

AI-Powered CRM Analytics & Offline Speech-to-Text Engine

Developed and deployed an intelligent CRM analytics pipeline for an enterprise-level application with over 5 million active users. The core engine features a highly optimized, fully offline Speech-to-Text (STT) model achieving an exceptional 3.5% Word Error Rate (WER). Once the audio is transcribed, a custom NLP pipeline performs deep behavioral analysis on customer support calls. It automatically scores agents (0-100) on soft skills, measures politeness, analyzes conversational traits (e.g., active listening, talkativeness), generates concise call summaries, and accurately categorizes call topics. This end-to-end AI solution significantly improved quality assurance, agent performance monitoring, and overall customer satisfaction in a high-volume production environment.

Intelligent RFM Customer Segmentation & Clustering

Engineered 'iRFM', an advanced ML-based customer segmentation system capable of processing hundreds of thousands of transaction records. Unlike traditional, static RFM (Recency, Frequency, Monetary) models, this engine utilizes machine learning clustering combined with custom marketing heuristics to dynamically score and categorize users. The algorithm automatically benchmarks individual user behavior against the broader customer base, delivering highly personalized, data-driven insights. Deployed in a top-tier financial and utility application (5M+ installs), the system directly empowered the marketing team to execute targeted promotional strategies, optimize budget allocation, and significantly boost user retention rates.

Multi-Horizon Forex Prediction & Market Sentiment AI

Developed a comprehensive machine learning architecture for multi-horizon financial forecasting (Short, Medium, and Long-term predictions). The model is trained on a dynamic dataset utilizing 15 complex macroeconomic features, including global/local gold prices, crude oil, and EUR rates. A key technical differentiator is the integrated custom web crawler that aggregates real-time financial news, applies NLP for sentiment analysis, and feeds the 'market pulse' directly into the predictive model as a dynamic feature. The system's performance is strictly evaluated using RMSE, MAE, MSE, and Directional Accuracy (DA) metrics. Deployed within a major financial services app, it successfully identifies market trends and provides actionable predictive insights in a highly volatile economic environment.

APIN-S-25-10316.pdf

HosseinJandaghi

Work experience

Education

AI-Powered CRM Analytics &amp; Offline Speech-to-Text Engine

Intelligent RFM Customer Segmentation &amp; Clustering

Multi-Horizon Forex Prediction &amp; Market Sentiment AI

Hossein
Jandaghi

AI-Powered CRM Analytics & Offline Speech-to-Text Engine

Intelligent RFM Customer Segmentation & Clustering

Multi-Horizon Forex Prediction & Market Sentiment AI