Abdellatif Dalab
Verified Expert in Engineering
Data Scientist and Software Developer
Riyadh, Riyadh Province, Saudi Arabia
Toptal member since December 8, 2021
Abdellatif builds ML systems where the hard problems lie not in the model itself but in the surrounding architecture: how inference scales under cost constraints, how search quality degrades across edge cases, and how features that work in a notebook survive production traffic. He has seven years of experience delivering AI features at different scales, specializing in LLMs, semantic search, and inference pipelines. Abdellatif thinks in systems, not tools.
Portfolio
Experience
- Python - 9 years
- Natural Language Processing (NLP) - 8 years
- Generative Pre-trained Transformers (GPT) - 7 years
- Hugging Face - 7 years
- Scikit-learn - 7 years
- PyTorch - 7 years
- SQL - 7 years
- Transformer Models - 5 years
Preferred Environment
Google Colaboratory (Colab), Visual Studio Code (VS Code), Jupyter Notebook, MacOS, GitHub
The most amazing...
...thing I’ve developed was a replacement SOTA deep learning server at a Toronto-based startup that led to an acquisition by Sprout Social.
Work Experience
Senior Machine Learning Engineer
https://abwab.ai/
- Worked on custom credit engines for SMEs in a forward-deployed model.
- Engineered credit risk signals for probability-of-default prediction.
- Analyzed financial data warehouses for feature extraction and evaluated different ML solutions.
Lead ML & DS Instructor
Ironhack
- Led instruction in production ML systems covering supervised/unsupervised learning, neural networks, model evaluation, and statistical modeling for a structured data science curriculum.
- Delivered applied modules on LLM systems, including retrieval-augmented generation (RAG), LLM fine-tuning, and prompt engineering for production use cases.
- Taught end-to-end ML engineering workflows covering Python, SQL, data pipelines, and model deployment practices.
Senior Applied ML Scientist
Sprout Social
- Led the design, research, and development of a large-scale LLM summarization system for real-time social data that utilizes optimized sampling, clustering, and summary generation.
- Launched an LLM-powered translation feature adopted by 2,000+ businesses with 500,000+ translations and 80% repeat usage.
- Released a post-generation optimization feature for multi-network publishing, reaching 30,000+ generations with 80% adoption.
- Directed AI feature centralization, building standardized monitoring protocols and an end-to-end analytics platform to track generative feature performance.
- Created a document retrieval evaluation framework with synthetic data, cross-domain testing, and hybrid search to measure context relevance and generalization.
- Oversaw sentiment analysis R&D and built a standardized hyperparameter tuning pipeline (RayTune, multiple architectures) as part of core DS workflows.
Senior Software Engineer, Machine Learning & NLP
Sprout Social
- Led R&D in large language model optimization for sentiment classification.
- Achieved a 30x throughput increase and 15x infrastructure cost reduction via knowledge distillation, quantization, and multilingual model centralization.
- Built internal fine-tuning and training pipelines for LLMs, including data generation for underrepresented languages.
- Deployed production ML services (sentiment, summarization) on centralized infrastructure.
Lead Machine Learning Engineer
Repustate
- Developed interpretable supervised and unsupervised deep learning solutions (BERT, custom attention layers, hierarchical attention networks), enabling scalable client-specific models without tagged data.
- Designed a new generation gRPC microservices API connecting Go applications with Python ML servers, reducing onboarding time from four weeks to 1–3 days.
- Built a multilingual transcription service replacing Amazon Transcribe, cutting annual expenses by 14x.
- Improved inference throughput 2–3x using ONNX quantization and batching.
- Partnered with sales and client teams, contributing to 10+ new client wins through technical leadership and custom ML solutions.
- Contributed to the startup, which was acquired by Sprout Social, and most of the accomplishments listed were integrated into the acquirer's ecosystem.
Data Scientist
Decathlon
- Built multiple AI/ML solutions across NLU, recommendation, forecasting, and computer vision to support Decathlon’s personalization and analytics strategy.
- Delivered major cost savings by developing in-house tools: a data-visualization pipeline (-$60,000/year) and an NLU system for reviews (-$15,000/year).
- Deployed production ML APIs, including a visual search engine, product-article recommendation system, and turnover forecasting models.
- Improved customer insights via unsupervised topic modeling and sentiment analysis (sentence transformers, GPT-2).
- Collaborated with MLOps to deploy TensorFlow models and mentored new DS hires.
- Contributed to earlier work, including recommendation systems with LSTMs, attention-based models, and computer vision for object detection.
Machine Learning Developer
Societe Generale
- Developed a BI reporting tool using MicroStrategy.
- Contributed to data visualization projects using Tableau.
- Helped develop a web application using the Django framework.
Experience
Adjacent Open Source Project
https://github.com/abdelatifsd/AdjacentEducation
Bachelor's Degree in Information Technology
Concordia University - Montreal, Quebec, Canada
Certifications
Improving Deep Neural Networks: Hyperparameter Tuning, Regularization, and Optimization
Coursera
Structuring Machine Learning Projects
Coursera
Neural Networks and Deep Learning
Coursera
Machine Learning with Python in Data Science
Udemy
Machine Learning
Stanford University | via Coursera
Skills
Libraries/APIs
Keras, Scikit-learn, NumPy, Pandas, PyTorch, XGBoost, TensorFlow, React, Protobuf, Stripe API, AsyncTask
Tools
Jenkins, Git, GitHub, Tableau, Open Neural Network Exchange (ONNX), PyPI
Languages
Python, SQL, HTML, CSS, JavaScript, Go
Platforms
Amazon Web Services (AWS), Google Cloud Platform (GCP), Visual Studio Code (VS Code), Jupyter Notebook, Docker, MacOS, Amazon EC2
Storage
Redshift, PostgreSQL, Amazon S3 (AWS S3), MySQL, Google Cloud Datastore, Database Programming, MongoDB, Data Pipelines, Neo4j
Frameworks
Flask, Django, gRPC
Paradigms
Database Design, Model Context Protocol (MCP)
Other
Artificial Intelligence (AI), Data Analysis, Machine Learning, Natural Language Processing (NLP), Generative Pre-trained Transformers (GPT), Data Science, AI Model Training, Google Data Studio, Deep Learning, OOP Designs, Data Structures, Algorithms, Programming Languages, Hugging Face, FastAPI, Large Language Models (LLMs), Transformer Models, Knowledge Distillation, Generative Artificial Intelligence (GenAI), Forecasting, Time Series Forecasting, Logistic Regression, APIs, Anthropic, Vector Databases, Statistical Analysis, Statistical Methods, MicroStrategy, Google Colaboratory (Colab), MLflow, Amazon Route 53, Fine-tuning, Semantic Web, Embeddings from Language Models (ELMo), RAG Pipelines, attention mechanisms, FAISS, model optimization, Multilingual Language Models (MLMs), Semantic Search, summarization systems, Centralized ML Infrastructure, Containerization, SQLModel, analytics systems, Clustering, Information Retrieval, Recommendation Systems, RAG Systems, Statistics, Probability Theory, OpenAI, Prompt Engineering, Retrieval-augmented Generation (RAG), Risk Modeling, Credit Default Swap (CDS)
How to Work with Toptal
Toptal matches you directly with global industry experts from our network in hours—not weeks or months.
Share your needs
Choose your talent
Start your risk-free talent trial
Top talent is in high demand.
Start hiring