Building AI-powered products
from research to full-stack deployment.

About

Production AI Engineer with 7+ years of engineering experience, including 3+ years specializing in AI/ML and 4+ years in full-stack software development. I build end-to-end AI systems, from large-scale data pipelines and model development to cloud deployment, MLOps, and production operations.

My career started in full-stack product engineering, where I built SaaS platforms, REST APIs, databases, and university-facing production systems. Through graduate study and applied research, I moved deeper into AI, recommender systems, NLP, explainability, and information quality. Today, I focus on bridging software engineering and machine learning by turning research ideas into reliable, observable, and maintainable AI-powered products.

Skills & Expertise

A practical mix of AI research, production engineering, cloud operations, product thinking, and startup execution. I focus on skills that help move ideas from prototypes to reliable systems people can actually use.

Projects

Selected product, research, and applied AI projects focused on production systems, recommender systems, LLM workflows, and trustworthy information quality.

Production Systems

PhotoCans - Web App

Ended Jul 2021

Co-founded and delivered the web platform for a privacy-first family photo management product, covering product pages, admin workflows, sales operations, customer onboarding, licensing, and support.

ASP.NET Core C# REST APIs SQL Server

Impact: Served 100+ users, implemented 33 REST endpoints and 12+ business workflows, and maintained multiple production releases.

Details

Problem Families needed a private way to organize, browse, and share memories across devices without depending on public cloud storage.

My Contribution Co-led product development, backend architecture, admin workflows, licensing flows, customer onboarding, and release support.

Engineering Focus Built API-backed workflows for users, products, purchases, licensing, profile management, and operational administration.

Watch Now

Hall Reservation Automation

Ended Feb 2019

Designed and developed ReserveAfzar for Bu-Ali Sina University as a web-based hall scheduling and reservation platform, enabling academic staff to book university halls online for events, seminars, and sessions.

ASP.NET C# SQL Server System Design

Impact: Deployed on the university's internal server, reducing manual booking time, improving hall availability visibility, and supporting operational reporting for administrators.

Details

Problem Hall reservations were handled manually, making availability, recurring bookings, and administrative reporting harder to manage.

My Contribution Gathered requirements from university staff, designed the reservation workflow, developed the web app, and supported adoption.

Engineering Focus Delivered calendar-based booking, hall availability views, recurring reservation support, and operator reporting.

MeydaNews

Present

Building an AI-powered media analysis platform for monitoring news sources using machine learning, social-media signals, and production-grade MLOps infrastructure on AWS.

MLOps AWS FastAPI Next.js

Impact: Designed an end-to-end ML lifecycle covering feature engineering, model training, validation, deployment, monitoring, and CI/CD.

Details

Problem News consumers and analysts need clearer signals about source credibility, framing, bias, and misinformation risk.

My Contribution Founded the product direction and built the AI-backed credibility analyzer, API workflow, interface, and deployment path.

Engineering Focus Connects ML scoring, source analysis, monitoring, production APIs, frontend presentation, and cloud operations.

Watch Now

Research and Development

Exogenous Cues to Information Quality

Ended Feb 2026

Built a reproducible research pipeline for engineering contextual signals that help evaluate information quality in algorithmic newsfeeds.

Feature Engineering Information Quality Evaluation

Impact: Converted large-scale social-media data into structured quality signals for downstream modeling and recommendation experiments.

GitHub

Quality-Aware News Source Recommender

Ended Feb 2026

Designed a recommender system that integrates 41 engineered information-quality signals from 5M+ social-media records to improve the reliability of recommended news sources.

Recommender Systems Ranking ML Pipelines AWS

Impact: Increased average recommended source quality by +3 points while maintaining baseline precision.

GitHub

Automated Reddit Scraper

Ended Feb 2026

Developed a scalable Reddit data collection pipeline for extracting threads, comments, interaction metadata, and research-ready datasets for political discourse analysis.

Data Pipeline APIs Research Data Validation

Impact: Supported reproducible data collection and quality checks for social-media and political news research.

A Framework for Generating Explainable Recommendations via Multi-Task Learning

Ended Sep 2023

Built a deep learning framework that jointly models recommendation accuracy and explanation generation using user preferences, review text, and neural text-generation methods.

Multi-task Learning NLP Explainable AI Recommenders

Impact: Evaluated on 151K reviews and ratings with strong rating prediction and explanation-generation results.

GitHub

Explainable Recommendation System using Reinforcement Learning

Ended Sep 2023

Explored reinforcement-learning strategies for improving personalized recommendation and explanation generation, with a focus on explainability, user preference modeling, and generated text quality.

Reinforcement Learning Explainable AI Text Generation Personalization

Impact: Investigated how optimization and generated explanations can improve transparency in recommendation workflows.

GitHub

Persian-Handwritten-Digit-Recognition

Academic project

Built a computer-vision project for recognizing Persian handwritten digits, using image preprocessing and a sequence-matching approach for digit classification.

Computer Vision OCR Image Processing Pattern Matching

Impact: Showcases applied ML and image-processing work for Persian digit recognition.

GitHub

AI-driven Tools

Chatbot Evaluation Framework

Ended Feb 2026

Developed a Python framework for benchmarking LLM and chatbot responses across prompts, reference outputs, quality criteria, and repeatable evaluation workflows.

LLM Evaluation Benchmarking Python Testing

Impact: Made conversational AI assessment more systematic, reproducible, and easier to compare across model outputs.

GitHub

Translation Quality Evaluation

Ended Jan 2026

Built an end-to-end translation pipeline for preprocessing, machine translation, post-processing, glossary checks, semantic similarity scoring, and quality-based publish/review decisions.

Machine Translation NLLB Quality Evaluation Automation

Impact: Automated translation validation so low-quality outputs can be flagged, retranslated, or routed for human review.

GitHub

DocPilot - AI Study Assistant

Ended Jan 2026

Built a production-style LLM backend integrating RAG, ReAct agent workflows, function calling, prompt orchestration, persistent memory, and multi-turn stateful interactions.

RAG ReAct Agents Function Calling FastAPI

Impact: Demonstrates grounded, tool-driven reasoning over study materials with structured backend workflows.

GitHub

Tech Stack

Languages

C#

Python

R

C++

JavaScript

Web & Frameworks

HTML

CSS

Bootstrap

Blazor

Next.js

.NET Core

FastAPI

AI / Data

TensorFlow

PyTorch

NumPy

Pandas

Scikit-learn

OpenCV

ML.NET

Jupyter

LangChain

Cloud / DevOps / Tools

Docker

Prometheus

AWS

Azure

MySQL

SQLite

MS SQL Server

Git

GitHub

Visual Studio

IIS

Linux

Publications

Gallery

A compact look at the human side of my work: presenting ideas, building in teams, and staying competitive beyond the keyboard.

Social Media for Democracy Project

A research presentation on using social-media signals to evaluate information quality in digital news environments.

Project Website

Recommender Systems Seminar (Farsi)

A virtual graduate seminar session in Farsi, showing my academic presentation style and ability to explain AI topics clearly.

Volleyball Team Captain

Leading on the court: teamwork, communication, and shared momentum in a very real form.

Office Christmas Moment

A small reminder that good teams are built through trust, warmth, and everyday shared rituals.

PES2019 Competition Runner-up

Second place in a PES2019 competition: strategy, focus, and competitive energy outside work.

Contact

Open to AI/ML engineering, applied research, and product-focused AI roles, especially where research needs to become reliable production software.

GitHub LinkedIn Instagram YouTube Email Stack Overflow Google Scholar