Ismail is available for hire
Hire IsmailIsmail Karchi
Verified Expert in Engineering
Data Scientist and Developer
Casablanca, Casablanca-Settat, Morocco
Toptal member since July 13, 2022
Bio
Ismail is a data scientist with over a decade of experience working on high-value projects in the eCommerce, contact center, agriculture, petroleum, and transportation sectors. He specializes in descriptive and prescriptive data-oriented solutions, from design to implementation. Besides his extensive work experience, Ismail has also published a book on machine learning and statistical analysis application in the oil storage process.
Portfolio
WhyHow.AI
Python, Knowledge Graphs, OpenAI, Graph Databases, Neo4j...
Self-employed
Python 3, Clustering, Visual Studio Code (VS Code), Jupyter Notebook, Slack...
DM Technologies GmbH
Python, Machine Learning, Predictive Modeling, Data Science, Data Analysis...
Experience
Availability
Part-time
Preferred Environment
Visual Studio Code (VS Code), Jupyter Notebook, Slack, Machine Learning, Artificial Intelligence (AI), Full-stack, Python, Data Science, SQL, Microsoft Power BI, JSON, Data Analytics, Predictive Analytics, Predictive Modeling, Project Management, Text Analytics, Proof of Concept (POC), Minimum Viable Product (MVP), ChatGPT, AI Programming, Image Generation, APIs, Natural Language Processing (NLP), Generative Pre-trained Transformers (GPT), Natural Language Understanding (NLU), PyTorch, GitHub, Matplotlib, Information Extraction, Docker, Data Engineering, Generative Pre-trained Transformer 4 (GPT-4), OpenAI GPT-3 API, Full-stack Development, Spark, Chatbots, Data Scientist, Generative Pre-trained Transformer 3 (GPT-3), Statistical Methods, Statistical Data Analysis, Mathematical Analysis, Language Models, Artificial General Intelligence (AGI), Architecture, Classification, Text Classification, Data Modeling, OpenAI GPT-4 API, Data Cleaning, Unstructured Data Analysis, Natural Language Queries, Computer Vision, Image Recognition, Convolutional Neural Networks (CNNs), Frameworks, Large Language Models (LLMs), Supervised Learning, Unsupervised Learning, LlamaIndex, Building Information Modeling (BIM), Ontologies, Big Data, Data Collection, Modeling, Sentiment Analysis, Vectorization, Semantic Search, Product Discovery, OpenAI API
The most amazing...
...solution I've developed is a shipping engine with a conversion rate optimizer for Jumia, which impacted 6.8 million customers in 2020.
Work Experience
AI Infrastructure Engineer
2024 - PRESENT
WhyHow.AI
- Developed a RAG chatbot for financial-legal queries on enterprise documents using Neo4j, Pinecone, and OpenAI LLMs via LangChain and Chainlit. Used Python, Jupyter notebooks, and AWS to deliver a highly effective solution that impressed the client.
- Developed several RAG-based chatbots across diverse industries, leveraging advanced ontologies, graph databases, and vector stores. Handled a variety of data sources like PDFs, websites, and text. Delivered highly effective solutions that impressed clients.
- Contributed to a platform, supported deterministic chunk access and multi-graph management, and proposed innovative approaches combining knowledge graphs and vector databases for better data retrieval and AI accuracy.
Technologies: Python, Knowledge Graphs, OpenAI, Graph Databases, Neo4j, Amazon Web Services (AWS), Pinecone, ChromaDB, Vector Stores, Ontologies, Big Data, Data Collection, Modeling, Vectorization, Semantic Search, Product Discovery, ETL, Data Annotation, Technical Consulting, Vector Databases, AI Consulting, AI Prompts, OpenAI API
Data and AI Expert
2012 - PRESENT
Self-employed
- Published a book specialized in optimizing the oil storage process using statistical analysis (ISBN 978-3-8417-9268-6).
- Developed a profiling and marketing recommendation engine for the Moroccan railway company.
- Built an automatic bidding system for hotels on the Trivago metasearch for an online travel agency. The system is data-driven based on hotel segmentation and doubled the ROAS on 80% of the 5,489 hotels listed on the metasearch.
Technologies: Python 3, Clustering, Visual Studio Code (VS Code), Jupyter Notebook, Slack, SQL, Microsoft Power BI, Pandas, Scikit-learn, NumPy, XGBoost, Regression, Artificial Intelligence (AI), Machine Learning Operations (MLOps), Machine Learning, Full-stack, Python, Data Science, Generative Pre-trained Transformers (GPT), Natural Language Processing (NLP), Computer Science, Google Analytics, eCommerce, Google BigQuery, BigQuery, Mathematics, Applied Mathematics, Amazon Web Services (AWS), Statistics, Team Leadership, JSON, Text Mining, Web Scraping, Data Analytics, Data Mining, Data Reporting, Data Visualization, Business Analysis, Azure, Dashboards, Predictive Analytics, Predictive Modeling, Natural Language Toolkit (NLTK), Advisory, Technology Consulting, Data Analysis, OpenAI, Generative Pre-trained Transformer 2 (GPT-2), Time Series Analysis, Business Intelligence (BI), Statistical Analysis, Marketplaces, Analytics, Project Management, Data Extraction, Deep Learning, Text Analytics, Proof of Concept (POC), Minimum Viable Product (MVP), AI Programming, APIs, Natural Language Understanding (NLU), PyTorch, GitHub, Matplotlib, Information Extraction, Amazon SageMaker, Docker, PostgreSQL, Data Engineering, Generative Pre-trained Transformer 4 (GPT-4), OpenAI GPT-3 API, Full-stack Development, Spark, Marketing, Data Scientist, Generative Pre-trained Transformer 3 (GPT-3), Statistical Methods, Statistical Data Analysis, Mathematical Analysis, Language Models, Artificial General Intelligence (AGI), Architecture, Classification, Text Classification, Data Modeling, OpenAI GPT-4 API, TensorFlow, Financial Modeling, Data Cleaning, Unstructured Data Analysis, Large Data Sets, Natural Language Queries, Computer Vision, Image Recognition, Convolutional Neural Networks (CNNs), Frameworks, Large Language Models (LLMs), Chatbot Conversation Design, Chatbots, Recommendation Systems, Regression Modeling, Forecasting, Quantitative Analysis, LangChain, Generative Artificial Intelligence (GenAI), AI Design, Call Centers, Retrieval-augmented Generation (RAG), Supervised Learning, Unsupervised Learning, LlamaIndex, Building Information Modeling (BIM), Knowledge Graphs, Neo4j, Graph Databases, Prompt Engineering, Leadership, Technical Leadership, Agile, Data Structures, Database Architecture, Startups, Ontologies, Big Data, Data Collection, Modeling, Sentiment Analysis, Vectorization, Semantic Search, Product Discovery, ETL, Data Annotation, Causal Inference, Technical Consulting, Vector Databases, AI Consulting, AI Prompts, OpenAI API
Data Scientist | Machine Learning Expert
2023 - 2023
DM Technologies GmbH
- Developed a predictive model to forecast when a customer would buy next (cadence-based).
- Built a churn model to detect when a customer would churn.
- Developed a cross-selling algorithm to serve cross-seeling items for clients.
Technologies: Python, Machine Learning, Predictive Modeling, Data Science, Data Analysis, Data Modeling, Frameworks, Artificial Intelligence (AI), Large Language Models (LLMs), Chatbot Conversation Design, Chatbots, Recommendation Systems, LangChain, Generative Artificial Intelligence (GenAI), AI Design, Retrieval-augmented Generation (RAG), Supervised Learning, Unsupervised Learning, LlamaIndex, Building Information Modeling (BIM), Leadership, Technical Leadership, Agile, Data Structures, Database Architecture, Startups, Ontologies, Big Data, Data Collection, Modeling, Product Discovery, ETL, Data Annotation, Causal Inference, Technical Consulting, Vector Databases, AI Consulting
Artificial Intelligence GPT/LLM Developer
2023 - 2023
Syntrillo, Inc
- Developed a sophisticated chatbot tailored for post-stroke survivors, enhancing user engagement and overall user experience beyond the scope of general-purpose ChatGPT models.
- Implemented an advanced AI model to substantially improve the chatbot's understanding of complex medical inquiries, leading to quicker and more accurate query resolution than conventional ChatGPT solutions.
- Collaborated closely with healthcare professionals to infuse expert knowledge into the chatbot, improving its credibility and making it a preferred tool for healthcare practitioners over standard ChatGPT.
Technologies: Artificial Intelligence (AI), Generative Pre-trained Transformers (GPT), Generative Pre-trained Transformer 3 (GPT-3), Classification, Text Classification, Data Modeling, OpenAI GPT-4 API, Data Cleaning, Unstructured Data Analysis, Natural Language Queries, Frameworks, Large Language Models (LLMs), Chatbot Conversation Design, Chatbots, Recommendation Systems, LangChain, Generative Artificial Intelligence (GenAI), AI Design, Retrieval-augmented Generation (RAG), Supervised Learning, Unsupervised Learning, Building Information Modeling (BIM), Knowledge Graphs, Neo4j, Graph Databases, Prompt Engineering, Leadership, Technical Leadership, Agile, Data Structures, Database Architecture, Startups, Ontologies, Big Data, Data Collection, Modeling, Vectorization, Semantic Search, Product Discovery, Data Annotation, Technical Consulting, Vector Databases, AI Consulting, AI Prompts, OpenAI API
Data Scientist
2018 - 2020
Jumia
- Developed a customer profiling system to be used by sales and operations teams. It impacted 6.8 million customers from seven segments in 2020 and enhanced the operation's delivery success rate from 54% to 76%.
- Designed, executed, and reported the A/B testing of the shipping fees over five geographies in Africa—Morocco, Egypt, Nigeria, Kenya, and the Ivory Coast. The result-driven recommendations have been used to enhance the platform conversion rate.
- Built a shipping fees engine that sets the shipping fees dynamically and in a data-driven way to optimize target KPIs, including the net merchandise value, item sold, and operational gross margin.
Technologies: Python 3, SQL, Microsoft Power BI, Pandas, Scikit-learn, NumPy, XGBoost, Visual Studio Code (VS Code), Jupyter Notebook, Slack, Clustering, Regression, Artificial Intelligence (AI), Machine Learning, Full-stack, Python, Data Science, Computer Science, Optimization, A/B Testing, Google Analytics, eCommerce, Google BigQuery, BigQuery, Mathematics, Applied Mathematics, Statistics, Team Leadership, JSON, Text Mining, Web Scraping, Data Analytics, Data Mining, Data Reporting, Data Visualization, Business Analysis, Dashboards, Predictive Analytics, Predictive Modeling, Advisory, Technology Consulting, Data Analysis, Time Series Analysis, Business Intelligence (BI), Statistical Analysis, Marketplaces, Analytics, Project Management, Data Extraction, Proof of Concept (POC), Minimum Viable Product (MVP), AI Programming, APIs, Generative Pre-trained Transformers (GPT), Natural Language Processing (NLP), Natural Language Understanding (NLU), GitHub, Matplotlib, Information Extraction, Docker, PostgreSQL, Data Engineering, Full-stack Development, Spark, Data Scientist, Generative Pre-trained Transformer 3 (GPT-3), Statistical Methods, Statistical Data Analysis, Mathematical Analysis, Language Models, Architecture, Classification, Text Classification, Data Modeling, Data Cleaning, Unstructured Data Analysis, Large Data Sets, Convolutional Neural Networks (CNNs), Frameworks, Chatbots, Recommendation Systems, Regression Modeling, Forecasting, Quantitative Analysis, Generative Artificial Intelligence (GenAI), AI Design, Deep Learning, Supervised Learning, Unsupervised Learning, Building Information Modeling (BIM), Leadership, Technical Leadership, Agile, Data Structures, Database Architecture, Startups, Ontologies, Big Data, Data Collection, Modeling, Sentiment Analysis, Product Discovery, ETL, Data Annotation, Causal Inference, Technical Consulting, AI Consulting
Data Scientist
2016 - 2018
El Joumani Group
- Developed a model that predicts citrus quality during production time with over 87% accuracy.
- Created a model to predict oil prices and derivatives, enhancing the total cost of ownership by 4%.
- Transformed the abovementioned models into solutions, leading the process from ideation to delivery, then proposed and sold them.
Technologies: Python 3, Jupyter Notebook, Regression, Clustering, Visual Studio Code (VS Code), SQL, Pandas, Scikit-learn, NumPy, XGBoost, Artificial Intelligence (AI), Machine Learning, Full-stack, Python, Data Science, Computer Science, Optimization, Mathematics, Applied Mathematics, Statistics, JSON, Text Mining, Web Scraping, Data Analytics, Data Mining, Data Reporting, Data Visualization, Business Analysis, Dashboards, Predictive Analytics, Predictive Modeling, Advisory, Data Analysis, Time Series Analysis, Statistical Analysis, Analytics, Project Management, C#, R, Proof of Concept (POC), Minimum Viable Product (MVP), AI Programming, APIs, Generative Pre-trained Transformers (GPT), Natural Language Processing (NLP), Natural Language Understanding (NLU), GitHub, Matplotlib, Data Engineering, Full-stack Development, Spark, Marketing, Data Scientist, Generative Pre-trained Transformer 3 (GPT-3), Statistical Methods, Statistical Data Analysis, Mathematical Analysis, Language Models, Architecture, Classification, Data Modeling, Financial Modeling, Data Cleaning, Unstructured Data Analysis, Frameworks, Recommendation Systems, Regression Modeling, Forecasting, Quantitative Analysis, Supervised Learning, Unsupervised Learning, Building Information Modeling (BIM), Leadership, Technical Leadership, Data Structures, Database Architecture, Ontologies, Big Data, Data Collection, Modeling, Product Discovery, ETL, Data Annotation, Causal Inference, Technical Consulting
Full-stack Developer
2013 - 2016
El Joumani Group
- Developed a transport management system specific to oil product transportation.
- Built and maintained different modules, such as billing and fixed asset management.
- Created an ERP, the Agricultural Management System, for farm management.
Technologies: SQL, DevExpress, Full-stack, Data Science, Computer Science, Optimization, Mathematics, Applied Mathematics, Statistics, JSON, Data Analytics, Business Analysis, Advisory, C#, Proof of Concept (POC), Minimum Viable Product (MVP), GitHub, Matplotlib, Data Engineering, Full-stack Development, Data Scientist, Architecture, Data Modeling, Financial Modeling, Data Cleaning, Supervised Learning, Leadership, Technical Leadership, Data Structures, Database Architecture, Big Data, Data Collection, Modeling, Product Discovery, ETL, Technical Consulting
Experience
Web Service to Predict Real-time Calls
Developed and implemented a web service that accurately predicts call arrival volumes. The web service was developed using Python 3, FastAPI, and Docker,
and the final model was developed using scikit-learn, NumPy, Pandas, and XGBoost. I leveraged my expertise in statistical modeling and data analysis to create a reliable and efficient solution that helped optimize call center operations. The solution resulted in a 17% increase in call center productivity and a 3% reduction in agent cost.
and the final model was developed using scikit-learn, NumPy, Pandas, and XGBoost. I leveraged my expertise in statistical modeling and data analysis to create a reliable and efficient solution that helped optimize call center operations. The solution resulted in a 17% increase in call center productivity and a 3% reduction in agent cost.
A/B Testing of the eCommerce Platform's Shipping Fees
https://www.jumia.comWORK DONE
• Designed and conducted successful A/B testing for shipping fees and developed and executed test plans using statistical analysis to measure the impact of different shipping fees on customer behavior (conversion rate, average order value).
• Collaborated closely with the product team to design and implement optimal shipping fee options. Utilized Excel and Python to analyze data and generate reports to identify key insights and trends. Presented findings and recommendations to the team and stakeholders.
• Demonstrated strong analytical and problem-solving skills, as well as proficiency in A/B testing, statistical analysis, Excel, and Python. This project was conducted for a major eCommerce client.
The project resulted in defining the shipping fees strategy and an operational level matrix to take into account per region (Morocco, Egypt, Nigeria, Kenya, and Ivory Coast), resulting in a 0.43% flat increase in the conversion rate and a 17% increase in the average order value (varies by region), resulting in the implementation of the optimal shipping fee structure and strategy.
• Designed and conducted successful A/B testing for shipping fees and developed and executed test plans using statistical analysis to measure the impact of different shipping fees on customer behavior (conversion rate, average order value).
• Collaborated closely with the product team to design and implement optimal shipping fee options. Utilized Excel and Python to analyze data and generate reports to identify key insights and trends. Presented findings and recommendations to the team and stakeholders.
• Demonstrated strong analytical and problem-solving skills, as well as proficiency in A/B testing, statistical analysis, Excel, and Python. This project was conducted for a major eCommerce client.
The project resulted in defining the shipping fees strategy and an operational level matrix to take into account per region (Morocco, Egypt, Nigeria, Kenya, and Ivory Coast), resulting in a 0.43% flat increase in the conversion rate and a 17% increase in the average order value (varies by region), resulting in the implementation of the optimal shipping fee structure and strategy.
Delivery Success Predictor
Developed and implemented an innovative solution using Python 3 and XGBoost that accurately predicted order delivery success, whether the customer would accept the delivery or not (cash-on-delivery). This solution included a highly optimized and fully-functional web service, developed using FastAPI and deployed using Docker that efficiently predicted order delivery success based on a variety of features such as order size, value, distance, and time of day, customer data. The solution contributed to an 8.3% (flat) increase in delivery success.
Customer Feedback Analytics
Designed and developed customer feedback analysis for a major French telecommunication company with a yearly revenue of over €2 billion. Prior to my involvement, business and data analysts were struggling to make actionable plans to enhance the customer experience as they spent a significant amount of time analyzing feedback data. I led the development and implementation of an innovative solution utilizing Spacy, Python 3, and SQL database that accurately classified feedback into business-specified categories. The optimized MS SQL database, web service, and Power BI dashboard streamlined the process, resulting in a remarkable 96% reduction in the budget dedicated to feedback analysis. Collaborated with a global team of data analysts and quality analysts and worked closely with the quality operation director of the contact center. My contribution achieved a remarkable 95% accuracy rate in feedback categorization, allowing the company to better understand its customers and make informed decisions to improve their experience.
Optimizing Citrus Quality Control During Production with Advanced Predictive Modeling Techniques
http://www.joumani.net/I created and implemented a predictive model that accurately forecasts citrus quality during production time, achieving an exceptional accuracy rate of over 87%. Employed advanced statistical techniques and machine learning algorithms to analyze massive data sets, including historical production data, environmental factors, and fruit characteristics. The algorithm was built using Python 3, scikit-learn, and XGBoost. The database was built using MS SQL Server and the visualization layer with Power BI. I contributed to the development of the model's architecture, including feature selection, data preprocessing, and model selection. Collaborated with cross-functional teams to integrate the model into the production system, resulting in improved efficiency and reduced waste. Leveraged expertise in data science, statistical analysis, and programming languages such as Python, R, and SQL. This project was carried out for a leading citrus company.
Predictive Model for Oil Prices
http://www.joumani.net/activities/produit-petroliers/I developed a predictive model to forecast oil prices and derivatives, resulting in a noteworthy 4% increase in the overall cost of ownership. I had to leverage my expertise in data analysis and proficiency in programming languages such as Python and R to collect and analyze large sets of historical oil price data using various statistical tools and techniques. By applying machine learning algorithms and regression models, I successfully identified key patterns and trends in the data, enabling accurate predictions of future oil prices and derivatives. This groundbreaking solution was implemented for a prominent energy client, ultimately contributing to their business growth and profitability.
Transport Management System
http://www.joumani.net/activities/transport/I designed and developed a customized transport management system (TMS) tailored to the unique needs of oil product transportation. I leveraged my expertise in software development and knowledge of the transportation industry and collaborated closely with the client to gain a thorough understanding of their business requirements and workflows. Using cutting-edge technologies and frameworks such as C#, DevExpress, and SQL Server, I developed a robust TMS solution that enabled the streamlined management of the transportation process, from order placement to delivery tracking. The system also included modules for real-time monitoring of vehicle and driver performance, enhancing operational efficiency and ensuring compliance with industry regulations. This innovative solution was implemented for a leading oil and gas company, significantly improving its logistics operations and driving increased profitability.
Agricultural Management System | An ERP Solution for Increased Productivity
http://www.joumani.net/activities/agriculture/I developed and designed the Agricultural Management System, an ERP solution that streamlined farm management processes and increased productivity. Utilized cutting-edge technologies such as C#, DevExpress, and MS SQL Server to create a user-friendly interface that allowed farmers to manage daily operations, track inventory, and monitor crop yields. I contributed to a 30% increase in overall efficiency by implementing automated reporting and data analysis features. I also collaborated with a team of developers and project managers to ensure timely delivery and successful implementation for clients in the agricultural industry.
Customer Segmentation
Customer segmentation that I implemented using the k-means algorithm in Python 3 and integrated it with a Microsoft SQL Server database to support CRM solutions, resulting in improved commercial and operational processes. I specifically contributed to the development and implementation of a cash-on-delivery privilege for select customers, which resulted in a 12% increase in successful delivery. I used my expertise in Python 3, NumPy, pandas, and k-means algorithms to develop and implement the solution.
Marketplace Vendor Churn Predictor
https://sellercenter.jumia.ma/I designed and developed a vendor churn predictor for the marketplace using Python 3 and XGBoost to identify vendors at risk of leaving the platform.
I implemented data pre-processing techniques, feature engineering, and model selection to achieve an accuracy of 85%.
The solution was developed using Python 3 and Jupyter Notebook for the exploration, and the model deployed as automated scripts that push the data to MS SQL Server Database. The data is then consumed by the company's CRM software.
The solution contributed to reducing vendor churn by 12% within the first quarter of deployment.
I implemented data pre-processing techniques, feature engineering, and model selection to achieve an accuracy of 85%.
The solution was developed using Python 3 and Jupyter Notebook for the exploration, and the model deployed as automated scripts that push the data to MS SQL Server Database. The data is then consumed by the company's CRM software.
The solution contributed to reducing vendor churn by 12% within the first quarter of deployment.
Marketplace Vendor Segmentation
https://sellercenter.jumia.ma/I implemented vendor segmentation using a K-means algorithm in Python 3 and integrated it with an MS SQL Server database to support procurement solutions, resulting in improved vendor management and cost optimization. I specifically contributed to the identification of key vendor characteristics and the development of a segmentation model based on those characteristics, which resulted in a 15% decrease in procurement costs.
My expertise in Python 3, Numpy, Pandas, and K-means algorithms was instrumental in developing and implementing the segmentation model. Additionally, I utilized my knowledge of SQL to integrate the model with the company's database, enabling seamless use by the procurement team.
My expertise in Python 3, Numpy, Pandas, and K-means algorithms was instrumental in developing and implementing the segmentation model. Additionally, I utilized my knowledge of SQL to integrate the model with the company's database, enabling seamless use by the procurement team.
Dynamic Shipping Fees Engine Leveraging A/B Testing and Statistical Analysis
https://www.jumia.comI developed and implemented a dynamic shipping fees engine that optimized target KPIs, including net merchandise value, items sold, and operational gross margin. I leveraged the A/B testing result to set shipping fees in near real-time, resulting in a significant increase in revenue and profitability. I utilized statistical analysis while leveraging the A/B testing result to drive the engine logic. I also collaborated closely with cross-functional teams, including product managers, engineers, and data scientists, to ensure the successful delivery of the project. As a result of my efforts, the shipping fees engine significantly improved the client company's bottom line and customer satisfaction.
NLP-based Hotel Review Analysis and Custom Classification Model
I categorized and analyzed hotel reviews using natural language processing techniques to improve customer satisfaction ratings. I utilized Python and various NLP libraries, including NLTK, spaCy, and scikit-learn, to preprocess and classify reviews based on sentiment and topic. I developed a custom classification model with an accuracy of 93% using a combination of Naive Bayes and Support Vector Machine algorithms.
I also created an interactive dashboard using Power BI to visualize review trends and identify areas for improvement. I then presented findings and recommendations to senior management, resulting in the implementation of targeted improvements to hotel amenities and services.
Skills utilized: Natural Language Processing, Python, NLTK, spaCy, scikit-learn, classification algorithms (Naive Bayes, SVM), and Power BI.
I also created an interactive dashboard using Power BI to visualize review trends and identify areas for improvement. I then presented findings and recommendations to senior management, resulting in the implementation of targeted improvements to hotel amenities and services.
Skills utilized: Natural Language Processing, Python, NLTK, spaCy, scikit-learn, classification algorithms (Naive Bayes, SVM), and Power BI.
BTC Predictor
I developed and implemented BTC Predictor, a machine learning model that accurately predicted the future prices of Bitcoin. I utilized the Python programming language, along with various data analysis and visualization tools such as NumPy, Pandas, and Matplotlib.
I conducted extensive research on historical Bitcoin prices and market trends to identify key indicators that affect the price of Bitcoin. I used this information to train the machine learning model, which was able to predict future prices with 73% accuracy. I contributed to the model's success by refining the feature selection process and optimizing the hyperparameters of the machine learning algorithms.
The BTC Predictor was created for a client in the finance industry who was looking for a reliable tool to help them make informed investment decisions in the cryptocurrency market. The model was able to provide valuable insights and helped the client achieve a significant increase in their return on investment.
I conducted extensive research on historical Bitcoin prices and market trends to identify key indicators that affect the price of Bitcoin. I used this information to train the machine learning model, which was able to predict future prices with 73% accuracy. I contributed to the model's success by refining the feature selection process and optimizing the hyperparameters of the machine learning algorithms.
The BTC Predictor was created for a client in the finance industry who was looking for a reliable tool to help them make informed investment decisions in the cryptocurrency market. The model was able to provide valuable insights and helped the client achieve a significant increase in their return on investment.
Vehicle Image Classification
I developed and implemented a vehicle image identification system using TensorFlow, Keras, a Siamese model, and YOLOv5 for the Moroccan highway company. I utilized the Siamese model to compare two images and identify if they were of the same vehicle. The system achieved an accuracy rate of 97% by fine-tuning the YOLOv5 model and optimizing the training data. I contributed to the quantifiable result by optimizing the data preprocessing pipeline and implementing a custom loss function. This project enabled the client company to streamline its vehicle identification process and ultimately led to increased efficiency and fraud prevention.
The model was deployed using TensorFlow serving.
The model was deployed using TensorFlow serving.
Geolocation-based Hub Detection
I detected and analyzed geo location data to drive a hub detection system for a transportation company that carried out over 80,000 daily deliveries. I spearheaded the development of an algorithm that accurately identified and clustered frequent delivery locations to optimize routes and expedite delivery times. I leveraged my expertise in Python and SQL to analyze and visualize data, facilitating the identification of optimal hub locations in more than 20 cities across five countries. The algorithm was built using Python 3, K-means, and a visualization layer with Power BI.
I implemented the hub detection system to create more efficient routes with fewer stops, which resulted in a 25% increase in on-time deliveries and a 15% reduction in fuel costs. My contribution to the development of the algorithm played a pivotal role in achieving this feat. By accurately detecting and clustering frequent delivery locations, we were able to optimize routes and expedite deliveries.
Overall, the hub detection system greatly improved the transportation company's delivery operations. Our unique skill in developing algorithms that could accurately identify and cluster frequent delivery locations proved to be invaluable in achieving these results
I implemented the hub detection system to create more efficient routes with fewer stops, which resulted in a 25% increase in on-time deliveries and a 15% reduction in fuel costs. My contribution to the development of the algorithm played a pivotal role in achieving this feat. By accurately detecting and clustering frequent delivery locations, we were able to optimize routes and expedite deliveries.
Overall, the hub detection system greatly improved the transportation company's delivery operations. Our unique skill in developing algorithms that could accurately identify and cluster frequent delivery locations proved to be invaluable in achieving these results
Profiling and Marketing Recommendation Engine
I crafted a cutting-edge profiling and marketing recommendation engine for the Moroccan railway company, facilitating personalized communication with customers and enhancing their experience.
I leveraged advanced machine learning algorithms and predictive analytics methodologies to analyze customer data and build unique customer profiles, enabling the railway company to tailor its marketing strategies to individual customers.
I implemented the engine using Python3, scikit-learn, Flask, and MySQL; Deployed with Docker and integrated it with the company's existing customer relationship management (CRM) system.
As a result of my contribution, the recommendation engine increased customer engagement by 35%, leading to a significant boost in revenue and customer satisfaction.
I leveraged advanced machine learning algorithms and predictive analytics methodologies to analyze customer data and build unique customer profiles, enabling the railway company to tailor its marketing strategies to individual customers.
I implemented the engine using Python3, scikit-learn, Flask, and MySQL; Deployed with Docker and integrated it with the company's existing customer relationship management (CRM) system.
As a result of my contribution, the recommendation engine increased customer engagement by 35%, leading to a significant boost in revenue and customer satisfaction.
Dialect Conversion Tool
I built a dialect conversion tool using GPT-3.5 Turbo. The tool was geared to translate from one English dialect to another in blogposts and localize them in the target dialect. The tool was built using Python 3, GTP-3.5 Turbo API, and Flask.
Education
2009 - 2012
Master's Degree in Mathematics and Computer Science
Mohammadia School of Engineers - Rabat, Morocco
Skills
Libraries/APIs
Pandas, Scikit-learn, NumPy, XGBoost, Matplotlib, OpenAI API, Natural Language Toolkit (NLTK), PyTorch, TensorFlow, SpaCy, OpenCV
Tools
Slack, Microsoft Power BI, ChatGPT, Amazon SageMaker, AI Prompts, Google Analytics, BigQuery, GitHub, C#.NET WinForms, You Only Look Once (YOLO)
Languages
Python 3, SQL, Python, C#, R, C#.NET
Paradigms
Business Intelligence (BI), Building Information Modeling (BIM), Agile, ETL, Siamese Neural Networks
Platforms
Visual Studio Code (VS Code), Jupyter Notebook, Amazon Web Services (AWS), Azure, Docker
Storage
Database Architecture, JSON, Neo4j, Graph Databases, PostgreSQL
Frameworks
Spark, LlamaIndex, .NET
Industry Expertise
Project Management, Marketing
Other
Regression, Clustering, Machine Learning, Artificial Intelligence (AI), Full-stack, Data Science, Natural Language Processing (NLP), Mathematics, Applied Mathematics, Statistics, Data Analytics, Data Mining, Data Reporting, Data Visualization, Business Analysis, Predictive Analytics, Predictive Modeling, Advisory, Data Analysis, OpenAI, Time Series Analysis, Statistical Analysis, Analytics, Text Analytics, Proof of Concept (POC), Minimum Viable Product (MVP), AI Programming, APIs, Natural Language Understanding (NLU), Generative Pre-trained Transformers (GPT), Forecasting, Data Scientist, Generative Pre-trained Transformer 3 (GPT-3), Statistical Methods, Statistical Data Analysis, Mathematical Analysis, Language Models, Architecture, Classification, Text Classification, Data Modeling, Data Cleaning, Unstructured Data Analysis, Large Data Sets, Frameworks, Large Language Models (LLMs), Chatbot Conversation Design, Regression Modeling, Quantitative Analysis, Generative Artificial Intelligence (GenAI), AI Design, Call Centers, Supervised Learning, Unsupervised Learning, Prompt Engineering, Leadership, Technical Leadership, Data Structures, Ontologies, Data Collection, Modeling, Vectorization, Semantic Search, Product Discovery, Data Annotation, Causal Inference, Technical Consulting, Vector Databases, AI Consulting, eCommerce, Team Leadership, Text Mining, Web Scraping, Dashboards, Technology Consulting, Marketplaces, Deep Learning, Information Extraction, Data Engineering, OpenAI GPT-3 API, Generative Pre-trained Transformer 4 (GPT-4), Chatbots, OpenAI GPT-4 API, Financial Modeling, Natural Language Queries, Computer Vision, Image Recognition, Convolutional Neural Networks (CNNs), Recommendation Systems, LangChain, Retrieval-augmented Generation (RAG), Knowledge Graphs, Product Management, Data Science Product Manager, Startups, Big Data, Sentiment Analysis, Computer Science, Optimization, Machine Learning Operations (MLOps), A/B Testing, Google BigQuery, Generative Pre-trained Transformer 2 (GPT-2), Data Extraction, Image Generation, Full-stack Development, DevExpress, CRM APIs, Algorithms, Simulations, Artificial General Intelligence (AGI), Pinecone, ChromaDB, Vector Stores, OCR
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