Mahmood Karimian

Currently working as a Senior Python AI Engineer on AI and GenAI-oriented software engineering projects within BNP Paribas. Responsibilities include designing and developing production-level backend components, APIs, and Python services to support AI-powered internal tools and workflows. Contributing to LLM-based features, automation workflows, data processing, PostgreSQL-based services, and integration with existing enterprise systems. Working in a production environment with Python, FastAPI, PostgreSQL, Kubernetes, ArgoCD, CI/CD practices, and internal platform constraints. Translating business needs into clear technical requirements and collaborating with technical and business stakeholders to deliver practical, reliable, and maintainable solutions. Building and improving prototypes, MVP-style components, and production-ready services with a focus on software quality, scalability, documentation, testing, refactoring, and business impact.

• Building an AI-powered B2B platform based on LLM-driven and agentic systems for research, orchestration, and automation. • Designing multi-agent workflows and reusable evaluation metrics to measure task completion, robustness, and efficiency. • Owning system architecture, evaluation strategy, and iterative deployment with early users.

As the Tech Lead and AI Expert at Decorer, I spearheaded the development of advanced AI models for image recognition and natural language processing, enabling users to input text descriptions or upload images to receive personalized decor recommendations. I led the creation of a robust data pipeline to curate and preprocess a high-quality decor dataset, optimized AI models for real-time performance, and collaborated with the design team to ensure seamless integration of AI-driven features into the app. My efforts enhanced user satisfaction by delivering highly accurate and context-aware recommendations, driving the success of this innovative AI-powered interior design platform.

As the Tech Lead and AI Expert at Tukki, an AI-powered travel planner, I developed intelligent systems that provided users with personalized travel destination recommendations based on preferences and behavior. I led the creation of robust data pipelines, implemented advanced NLP and recommendation models, and optimized the platform for seamless user experience, enhancing its ability to help users discover their next trip effortlessly.

Work on Data Stream mining and Transfer Learning

Dideo - Ultimate Video Search Engine Dideo is a powerful meta video search engine with access to more than 5 billion videos on the web. More than 8 million users watch at least 25 million videos per month on this platform. Responsibilities for this service: - AI Team Lead - Analyzing and handling a substantial amount of real-time data - Designing and implementing a parental safety service in compliance with policies - Conducting data stream mining and anomaly detection to combat fake traffic Technologies include: - Python and its ML libraries for analyzing and retrieving desired content - ElasticSearch for storing metadata and recommended systems - Redis as a versatile data store for real-time processes Achievements include: - Development of a video meta-search service available on websites, applications, and web browser extensions - Implementation of content analysis and data stream mining services - Introduction of an AI-powered advertisement service

As the Tech Team Lead for TimeShark, I drive the development and deployment of cutting-edge AI solutions that redefine call answering for businesses. Our mission, "The Future of Call Answering – Powered by AI," focuses on delivering an exceptional caller experience through our AI receptionist tailored for restaurants and businesses. I lead a talented team in designing and implementing intelligent systems that handle calls seamlessly, ensuring accuracy, efficiency, and a personalized touch. By combining advanced natural language processing and AI-driven automation, we empower businesses to enhance customer engagement, streamline operations, and provide unmatched service quality.

Dolphin creates software products and services powered by Artificial Intelligence to help companies and developers provide a more incredible experience for their end-users and their employees. Our team emphasizes clear, assertive communication that empowers us to work in an efficient, simple organization. The team is led by software engineers and data scientists with exceptional experience making high-quality, large-scale software products. We intend to provide a stimulating work environment with good opportunities for advancement to motivate the happiest employees.

Teaching, supervising, and mentoring 20 top students from all over the country. The goal of the program is to train elite people as machine learning engineers.

Designing and implementing a multi-domain text-based dialogue system, Working as the Senior Machine Learning Engineer, and the head of the R&D team. • Implementing a Natural and Spoken Language Understanding system • Building a closed-domain language model for Informal texts • Development of a closed-domain spell-corrector for Informal texts • Implementing the answer generation module • Building a statistical learning based module for sequence labeling • Deploying the infrastructure of the whole project with Java

Designing tools and platforms for trading in stock and OTC markets, with a focus on optimizing database transactions. Responsibilities: • Research and development for optimizing database transactions. • Designing solutions to enhance the utilization of SQL Server databases. Key Technologies: • Microsoft SQL Server • Git Achievements: • Significantly accelerated SQL transaction speed, achieving a 10x improvement.

Working on search engine and responsible for text mining and text preprocessing • Developing a search engine as a member of the R&D team • Natural Language Processing • Text preprocessing (tokenization, normalization, stemming) Technologies include: • Java, Git • Apache Lucene Achievements include: • Research and development for text preprocessing • Introduction of Apache Lucene in project

Research Assistance - Working on some papers on Machine Learning and Video Classification