Belal is available for hire
Hire BelalBelal Esawe
Verified Expert in Engineering
Machine Learning Engineer and Developer
Vancouver, BC, Canada
Toptal member since June 21, 2023
Bio
Belal is a senior machine learning (ML) engineer with over a decade of experience developing and deploying AI-driven systems. His work has spanned LLMs, multi-agent frameworks, classification, recommendation systems, and computer vision. He has been involved across the full ML lifecycle, including research, model development, deployment, and system optimization. Belal is experienced with technologies such as Claude Code, GPT, LangChain, LangGraph, LLama, AWS, NLP, and computer vision.
Portfolio
The Reynolds and Reynolds Company - Main
Machine Learning, Natural Language Processing (NLP), Python...
Modelcode AI
Agentic AI, Agentic Frameworks, Agile, AI Programming...
Jottix Inc.
Artificial Intelligence (AI), OpenAI, OpenAI o1, OpenAI GPT-4 API...
Experience
- Model Tuning - 8 years
- Prompt Engineering - 8 years
- Machine Learning - 8 years
- Python - 6 years
- OpenAI - 3 years
- Amazon Web Services (AWS) - 3 years
- Natural Language Processing (NLP) - 3 years
- DeepSeek - 1 year
Preferred Environment
Agentic AI, Amazon Web Services (AWS), Agentic Frameworks, APIs, Chatbots, Claude, ClickHouse, MySQL, Prompt Engineering, Model Tuning
The most amazing...
...thing I've built and deployed is a multi-agent AI lead engine that delivered human-like customer service and smarter automation at scale.
Work Experience
AI/ML Technical Lead - GenAI
2023 - PRESENT
The Reynolds and Reynolds Company - Main
- Developed and deployed an AI-driven lead engine system leveraging LangChain and LLM to build multi-agent framework to provide human-like personalized customer service.
- Designed a multi-agent framework for task automation, integrating tools for scheduling and real-time customer data retrieval.
- Developed an advanced RAG system customized to the client’s documentation.
- Built and integrated multiple MCP tools to enhance agent performance.
- Enhanced platform efficiency through advanced prompt engineering strategies.
- Collaborated cross-functionally with stakeholders to define requirements, ensuring product alignment with client goals.
Technologies: Machine Learning, Natural Language Processing (NLP), Python, Generative Pre-trained Transformers (GPT), Machine Learning Operations (MLOps), SpaCy, Natural Language Toolkit (NLTK), Redis, MySQL, New Relic, ClickHouse, ChatGPT, Large Language Models (LLMs), OpenAI, Artificial Intelligence (AI), LangChain, Generative Artificial Intelligence (GenAI), Hugging Face, OpenAI GPT-3 API, DeepSeek, Agentic AI, Conversational AI, Retrieval-augmented Generation (RAG), OpenAI API, LangGraph, API Integration, System Architecture Design, Fine-tuning, Open-source LLMs, Agile, Scrum, Neural Networks, Claude, Cursor AI, Agentic Frameworks, RAG Architecture, Anthropic, Workflow Automation, AI Integration, Vector Databases, OAuth, AI Architecture, AI Agent Orchestration, Agentic AI Systems, Claude Code, AI Agents, Codex, Claude API, Claude Agent SDK, AI Hallucinations Management, Agentic RAG Systems, AI Voice Agents, AI Engineering, LLM Integration, Speech-to-Text (STT), Text-to-Speech (TTS), AI Automation, FastAPI, REST APIs, Multi-agent Systems, Real-time Systems, Workflow Automation & System Integration
Senior Machine Learning Engineer - GenAI
2025 - 2026
Modelcode AI
- Built Morph, a self-serve code modernization platform that modernizes and migrates entire codebases across programming languages and frameworks using LLM agents.
- Integrated Claude Code API and productionized reliable tool-calling and structured outputs for modernization workflows.
- Designed a multi-agent pipeline for large-scale refactors and repo-wide transformations.
- Built and integrated multiple MCP tools for repository navigation, code edits, and automated build, run, and test execution to improve correctness.
- Developed an advanced RAG system customized to the client’s repositories to retrieve architecture-aware context, including entry points, dependencies, and module boundaries, and reduce hallucinations.
- Collaborated cross-functionally with stakeholders to define requirements, ensuring product alignment with client goals.
Technologies: Agentic AI, Agentic Frameworks, Agile, AI Programming, Amazon Web Services (AWS), API Integration, Artificial Intelligence (AI), Chatbots, CI/CD Pipelines, Claude, ClickHouse, APIs, Anthropic, Workflow Automation, TypeScript, AI Integration, Enterprise SaaS, Vector Databases, JavaScript, OAuth, AI Architecture, AI Agent Orchestration, Agentic AI Systems, Claude Code, AI Agents, Codex, Claude API, Claude Agent SDK, AI Hallucinations Management, Agentic RAG Systems, AI Engineering, LLM Integration, AI Automation, FastAPI, REST APIs, Multi-agent Systems, Real-time Systems, Workflow Automation & System Integration
AI Specialist
2024 - 2025
Jottix Inc.
- Developed a multi-agent workflow powered by LLM to create a financial advisor. This system is capable of providing personalized financial planning based on customer profile data.
- Created new tools to give the agents access to financial documentation to provide personalized financial planning.
- Created another agent to send personalized financial planning via email.
Technologies: Artificial Intelligence (AI), OpenAI, OpenAI o1, OpenAI GPT-4 API, Agentic Frameworks, RAG Architecture, AI Architecture, AI Agent Orchestration, Agentic AI Systems, Claude Code, AI Agents, Codex, Claude API, Claude Agent SDK, AI Hallucinations Management, AI Engineering, LLM Integration, AI Automation, REST APIs, Multi-agent Systems, Real-time Systems, Workflow Automation & System Integration
AI Specialist
2024 - 2024
Greenspace Mental Health Ltd.
- Served as the AI subject matter expert for a Canada-based mental health technology company, advising leadership and engineering teams on AI roadmap decisions, recommendation engine strategy, and model adoption best practices.
- Assessed the company’s data landscape and identified the most promising inputs for recommendation use cases, balancing predictive value, implementation complexity, and privacy considerations.
- Directed build-versus-buy analysis for ML capabilities, helping the client choose between custom models and pre-trained solutions based on product fit, scalability, compliance risk, and total cost of ownership.
- Shaped the technical foundation for future AI features by recommending production-ready approaches for data preparation, model evaluation, deployment, and ongoing performance monitoring.
- Advised on safe and effective AI adoption in a sensitive healthcare context, emphasizing human-in-the-loop design, transparency, and measurable product impact.
Technologies: Artificial Intelligence (AI), Machine Learning, Python, PyTorch, Large Language Models (LLMs), OpenAI GPT-3 API, OpenAI GPT-4 API, Recommendation Systems, Natural Language Processing (NLP), Claude API, Claude Agent SDK, AI Hallucinations Management, Agentic RAG Systems, AI Engineering, LLM Integration, AI Automation, Multi-agent Systems, Real-time Systems, Workflow Automation & System Integration
AI Lead
2024 - 2024
Konner Frey
- Participated in the product discovery phase to address all product requirements.
- Validated the idea by evaluating all challenges and bottlenecks the team could face during implementation.
- Led the product road mapping to define all necessary tasks to complete the MVP.
Technologies: Artificial Intelligence (AI), Natural Language Processing (NLP), Machine Learning, Deep Learning, Computer Vision, Distributed Computing, Agile, Agentic Frameworks, RAG Architecture, AI Architecture, Claude API, Claude Agent SDK, AI Hallucinations Management, AI Engineering, AI Automation, Multi-agent Systems, Workflow Automation & System Integration
Lead Machine Learning Engineer
2020 - 2023
XGen
- Developed ML-driven products that resulted in a provable 5-22% revenue lift for clients.
- Created recommendation engines delivering personalized products, driving increased sales and customer satisfaction.
- Implemented innovative solutions leading to 250 million weekly site visitor predictions.
- Transformed data insights into actionable strategies, resulting in improved conversion rates.
- Collaborated with cross-functional teams to identify business needs and align ML solutions accordingly.
- Streamlined processes, reducing time-to-market for new product features.
- Led research and development (R&D) efforts, staying ahead of industry trends and delivering cutting-edge ML systems.
- Communicated results and achievements to stakeholders, building trust and inspiring confidence.
- Mentored team members, fostering their growth and driving impactful contributions.
Technologies: Python, Amazon Web Services (AWS), Google Cloud, Machine Learning, Computer Vision, CI/CD Pipelines, Natural Language Processing (NLP), TensorFlow, Keras, Convolutional Neural Networks (CNNs), Supervised Machine Learning, Machine Learning Automation, PyTorch, Artificial Intelligence (AI), Videos, OpenCV, Pandas, Deep Learning, APIs, Natural Language Toolkit (NLTK), SpaCy, AI Programming, OpenAI GPT-4 API, Data Science, Generative Pre-trained Transformers (GPT), Machine Learning Operations (MLOps), Recommendation Systems, Image Generation, Generative Artificial Intelligence (GenAI), Image Processing, Data Pipelines, Google Vision API, Fashion, API Integration, Agile, Scrum, Neural Networks, AI Integration, Enterprise SaaS, AI Architecture, Computer Vision Algorithms, Residual Neural Networks (ResNets), AI Automation, REST APIs, Real-time Systems, Workflow Automation & System Integration
Lead Machine Learning Engineer
2020 - 2022
Niricson
- Developed an AI-powered product leveraging data fusion techniques to automate mapping and quantification of damage on concrete structures, enabling faster and more accurate inspections.
- Implemented advanced algorithms that significantly reduced inspection time and costs for infrastructure owners while maintaining high accuracy levels.
- Collaborated with cross-functional teams to gather and integrate diverse data sources, optimizing the performance and reliability of the damage detection system.
- Validated the product's effectiveness through extensive testing and evaluation, achieving a high accuracy rate.
- Streamlined the data processing pipeline, enabling seamless integration with existing infrastructure inspection workflows and maximizing operational efficiency.
- Provided technical expertise and guidance to clients, facilitating the successful implementation and utilization of the AI-based damage assessment solution.
- Conducted thorough analysis of inspection results, generating comprehensive reports and visualizations to aid decision-making and prioritize maintenance efforts.
- Monitored and incorporated user feedback, continuously enhancing the product's capabilities and user experience.
- Presented the product's success and benefits to infrastructure owners and industry professionals, fostering strong partnerships and driving adoption across the sector.
- Contributed to the advancement of the field by publishing research findings and presenting at conferences, solidifying my reputation as a thought leader in AI-driven infrastructure inspection.
Technologies: Python, Amazon Web Services (AWS), Google Cloud, Machine Learning, Computer Vision, QGIS, MATLAB, TensorFlow, Keras, Convolutional Neural Networks (CNNs), Supervised Machine Learning, Machine Learning Automation, PyTorch, Artificial Intelligence (AI), OpenCV, Pandas, Deep Learning, AI Programming, Data Science, Image Generation, Generative Artificial Intelligence (GenAI), Image Processing, Cloud Point, Data Pipelines, API Integration, Agile, Scrum, Neural Networks, AI Integration, Enterprise SaaS, AI Architecture, Computer Vision Algorithms, Residual Neural Networks (ResNets), AI Automation, REST APIs, Workflow Automation & System Integration
Machine Learning Engineer
2017 - 2020
PhotoSat
- Spearheaded the development and implementation of an AI-based mineral alteration mapping product, resulting in increased sales and revenue generation for the company.
- Streamlined the process of mineral alteration mapping using AI techniques, significantly reducing project timelines and enabling faster decision-making for clients.
- Positioned the AI product as a key driver of business growth, attracting new customers and expanding market reach.
- Demonstrated the value and impact of the AI solution through data-driven results, leading to improved client satisfaction and increased repeat business.
- Transformed the way geospatial data is analyzed by leveraging AI technology, driving efficiency and accuracy in mineral exploration and resource extraction projects.
- Conducted in-depth analysis of sales data and customer feedback to identify areas for product enhancement and revenue optimization.
- Established PhotoSat as a leader in AI-driven mineral alteration mapping, contributing to the company's reputation for delivering fast, accurate, and revenue-generating geospatial solutions.
Technologies: Python, Remote Sensing, QGIS, TensorFlow, Keras, Convolutional Neural Networks (CNNs), Supervised Machine Learning, Machine Learning Automation, PyTorch, Artificial Intelligence (AI), OpenCV, Pandas, Deep Learning, AI Programming, Data Science, Image Generation, Image Processing, Remini, Agile, Scrum, Neural Networks, AI Architecture, Computer Vision Algorithms, Residual Neural Networks (ResNets), AI Automation, Workflow Automation & System Integration
Experience
Multi-Agent Code Modernization Platform
https://modelcode.aiMorph is a self-serve, AI-powered code modernization platform that transforms large codebases from one language or framework to another using a multi-agent LLM architecture. It combines Claude Code API–based code generation with MCP tool integrations for repository understanding, code editing, and automated build and test execution, while using architecture-aware RAG to provide reliable context from the source codebase. The system is built to support large-scale refactoring with verification and functional testing steps that help ensure correctness, code quality, and minimal regression risk.
Multi-agent Workflow for Lead Generation
Context-Aware Agent Assistant is an LLM-powered customer service and lead-generation chatbot that I built at Gubagoo to automate high-quality customer interactions while improving lead conversion. It used a multi-agent LangChain-based architecture with prompt and tool routing, RAG over business documentation, and third-party tool integrations to handle context-aware conversations, qualify leads professionally, and support actions such as booking service appointments. The system was designed as a production-ready, API-first solution and was optimized for both response quality and cost efficiency.
Recommendation System for eCommerce Platforms
An eCommerce personalized recommendation system that, leveraging historical user interactions, processes vast data to deliver tailored recommendations. Through advanced algorithms, it analyzes purchase history, browsing habits, and demographics to generate accurate product suggestions, enhance customer satisfaction, drive sales, and foster loyalty with personalized recommendations that resonate. This system elevates eCommerce platforms, provides a dynamic shopping experience, and unlocks new growth opportunities in the competitive market.
Automatic Product Tagging
An automatic product tagging solution that uses advanced AI models to extract main categories and attributes from any product catalog, eliminating the need for manual tagging. This innovative solution automatically generates accurate tags, simplifying catalog management, saving valuable time, and enhancing efficiency and organization. It is a cutting-edge solution that revolutionized how product tagging is handled by replacing manual efforts with seamless automation and transforming the workflow of product categorization.
AI-powered Smart Product Search
An innovative product designed for eCommerce customers. Powered by machine learning, this system revolutionizes the shopping experience by seamlessly scanning outfit images and conducting a product search. It intelligently identifies all items within the image and presents them in a captivating carousel called Complete-the-look. This way, customers can effortlessly explore and purchase the entire outfit, enhancing convenience and style cohesion. Embracing the future of fashion shopping, where visual inspiration transforms into a seamless purchase journey, this system elevates eCommerce platforms by offering a curated shopping experience that drives conversion.
Remote Sensing Applications for Geospatial Analysis using AI
A system that uses AI models to detect water bodies and produce accurate digital elevation maps (DEM). This solution streamlines the process by automating waterbody detection in satellite imagery, allowing geospatial analysts to focus on data analysis. By precisely zero-leveling waterbodies, the project enhances DEM accuracy, benefiting hydrology, environmental modeling, and land management applications. I revolutionized geospatial analysis with this innovative project, optimizing workflows and delivering valuable insights for informed decision-making.
Satellite Imagery Analysis using AI for Mineral Alteration Mapping
This innovative solution utilizes AI models to analyze multispectral satellite images, enabling spectral matching and precise mineral detection. By harnessing data from satellites such as WorldView-3 and ASTER, this game-changing project for professional geologists in mining unveils unique spectral signatures of various minerals.
By leveraging rich multispectral data, Mineral Alteration Mapping aids exploration, resource estimation, and mining planning. With this system, geologists can efficiently detect and map mineral alteration, significantly reducing time and costs in the industry. By automating the process, advanced AI models analyze satellite images, providing invaluable insights into mineral presence and distribution, enabling geologists to focus on in-depth analysis and decision-making, thereby bolstering productivity and accuracy.
Ultimately, Mineral Alteration Mapping optimizes mineral exploration and mining operations, empowering geologists with an efficient tool to identify and map minerals, unlocking new discoveries and resource-management opportunities.
By leveraging rich multispectral data, Mineral Alteration Mapping aids exploration, resource estimation, and mining planning. With this system, geologists can efficiently detect and map mineral alteration, significantly reducing time and costs in the industry. By automating the process, advanced AI models analyze satellite images, providing invaluable insights into mineral presence and distribution, enabling geologists to focus on in-depth analysis and decision-making, thereby bolstering productivity and accuracy.
Ultimately, Mineral Alteration Mapping optimizes mineral exploration and mining operations, empowering geologists with an efficient tool to identify and map minerals, unlocking new discoveries and resource-management opportunities.
Craft Your own Game Character using Gen-AI
I employed stable diffusion and Null-Text Inversion techniques to generate and/or locally edit characters. Furthermore, I leveraged a fine-tuned language model for character dialogues to create a chat style that aligns with the selected character. By utilizing these advanced methodologies, I achieved impressive results and delivered high-quality outcomes for my client. My approach not only demonstrates technical proficiency but also emphasizes the ability to create engaging and dynamic content.
Defects Mapping in Infrastructure using AI
In this project, I focused on defect mapping in infrastructure. Leveraging AI, I developed a robust system capable of mapping a range of defects, including cracks, spalling, efflorescence, and joint failures. This model has been trained on a vast amount of data collected from diverse assets worldwide, ensuring its adaptability to different factors such as construction material, time of day, weather, and shadows.
By leveraging AI, this system can automate the identification and quantification of defects in infrastructure assets such as dams, chimneys, roads, and buildings. Accurately mapping defects is crucial to ensuring these assets' safety and structural integrity while simultaneously reducing the time and cost of manual investigations.
Through this project, I provided customers with valuable insights by quantifying defects and sorting them by severity. On the other hand, by presenting defect information in a systematic and organized manner and evaluating and prioritizing defects, I empowered engineers to efficiently assess the condition of infrastructure assets and take appropriate actions.
By leveraging AI, this system can automate the identification and quantification of defects in infrastructure assets such as dams, chimneys, roads, and buildings. Accurately mapping defects is crucial to ensuring these assets' safety and structural integrity while simultaneously reducing the time and cost of manual investigations.
Through this project, I provided customers with valuable insights by quantifying defects and sorting them by severity. On the other hand, by presenting defect information in a systematic and organized manner and evaluating and prioritizing defects, I empowered engineers to efficiently assess the condition of infrastructure assets and take appropriate actions.
Multi-agent AI System for Personalized Financial Planning
I delivered an end-to-end LLM-powered multi-agent financial advisory platform that generates personalized financial plans from customer profile data. I enhanced recommendation quality by building custom tools that allow agents to access relevant financial documentation and automate client communication through a dedicated email agent that sends tailored financial planning reports.
Education
2014 - 2020
PhD in Computer Engineering
University of Victoria - Victoria, BC, Canada
2012 - 2014
Master's Degree in Science
University of Northern British Columbia - Prince George, BC, Canada
Skills
Libraries/APIs
TensorFlow, Keras, PyTorch, OpenCV, Pandas, OpenAI API, Claude API, REST APIs, Natural Language Toolkit (NLTK), SpaCy, Google Vision API
Tools
Slack, Zoom, ChatGPT, Claude, Claude Code, Codex, Claude Agent SDK, MATLAB, Remini, DeepSeek, OpenAI o1
Languages
Python, TypeScript, JavaScript
Frameworks
LangGraph, Agentic Frameworks, Flask
Paradigms
Agile, Scrum, Real-time Systems, Distributed Computing, Model Context Protocol (MCP), Unit Testing
Platforms
Linux, Visual Studio Code (VS Code), Amazon Web Services (AWS), New Relic
Storage
Data Pipelines, Google Cloud, Redis, MySQL, ClickHouse, PostgreSQL
Other
Machine Learning, Computer Vision, Research, Complex Problem Solving, CI/CD Pipelines, Natural Language Processing (NLP), Convolutional Neural Networks (CNNs), Supervised Machine Learning, Machine Learning Automation, Artificial Intelligence (AI), Deep Learning, Chatbots, AI Programming, Data Science, Large Language Models (LLMs), OpenAI, LangChain, Image Generation, Generative Artificial Intelligence (GenAI), OpenAI GPT-3 API, Image Processing, Prompt Engineering, Model Tuning, Agentic AI, Conversational AI, Retrieval-augmented Generation (RAG), Fashion, API Integration, Fine-tuning, Open-source LLMs, Neural Networks, Cursor AI, RAG Architecture, Anthropic, Workflow Automation, AI Integration, Enterprise SaaS, Vector Databases, OAuth, AI Architecture, AI Agent Orchestration, Agentic AI Systems, AI Agents, AI Hallucinations Management, Agentic RAG Systems, AI Voice Agents, AI Engineering, Computer Vision Algorithms, Residual Neural Networks (ResNets), LLM Integration, AI Automation, FastAPI, Multi-agent Systems, Workflow Automation & System Integration, QGIS, Remote Sensing, Videos, APIs, OpenAI GPT-4 API, Generative Pre-trained Transformers (GPT), Machine Learning Operations (MLOps), Recommendation Systems, Hugging Face, LoRa, Cloud Point, System Architecture Design, Speech-to-Text (STT), Text-to-Speech (TTS), Axiom, Linear, Finite-state Machines (FSM), RAG Systems
Collaboration That Works
How to Work with Toptal
Toptal matches you directly with global industry experts from our network in hours—not weeks or months.
1
Share your needs
Discuss your requirements and refine your scope in a call with a Toptal domain expert.
2
Choose your talent
Get a short list of expertly matched talent within 24 hours to review, interview, and choose from.
3
Start your risk-free talent trial
Work with your chosen talent on a trial basis for up to two weeks. Pay only if you decide to hire them.
Top talent is in high demand.
Start hiring