Engineering that ships and scales

An AI system that reads and reasons over charts and graphs the way a human analyst would — answering complex questions about visual data automatically.

Architected a production-grade multimodal reasoning pipeline combining Qwen2.5-VL, OCR, and Zero-Shot Chain-of-Thought prompting. Achieved 93.96% accuracy on the ChartQA benchmark — outperforming state-of-the-art 72B parameter models at a tenth of the compute cost on an RTX 3090. Published at IEEE CAI 2026, sponsored by Microsoft.

A smart cataloguing system that automatically reads, tags, and organises hundreds of thousands of documents in a university's digital library — no manual data entry required.

Designed and deployed a scalable data ingestion and tagging pipeline for a major university's digital library. Integrated Florence-2, Tesseract OCR, YOLO, and NLP tooling to automate metadata extraction, visual classification, and semantic indexing across hundreds of thousands of documents.

A mobile app that identifies ingredients from a photo of your fridge and instantly generates personalised recipes — like having a private chef powered by AI.

Built a full-stack mobile application that combines Gemini API for intelligent recipe generation with a YOLO-powered computer vision layer that automatically detects ingredients from photos. The system delivers personalized, context-aware culinary recommendations in real time.

An AI platform that tells farmers exactly when and how much to water their crops, cutting water waste while keeping yields high.

Engineered an AI-driven precision agriculture platform using ensemble machine learning models to predict soil moisture and optimize irrigation scheduling. Reduced water usage while maintaining crop yield through predictive water management and sensor fusion.

A forecasting system that predicts how much cash each ATM will need and when, so a bank never runs out or overstocks — saving millions in operational costs.

Delivered an end-to-end ETL and forecasting system for one of Egypt's largest banking institutions. Combined Prophet and SARIMAX time-series models to predict ATM cash fluctuations, directly reducing operational costs and replenishment cycles. Ranked #1 out of 1,000+ interns for project impact.

A financial forecasting engine that predicts stock price volatility and quantifies uncertainty — giving traders a data-driven edge across different market conditions.

Built a financial forecasting engine that fuses LSTM and Transformer architectures with multi-head attention mechanisms and Quantile Regression to predict price volatility. Integrated TA-Lib for technical indicator preprocessing, enabling robust uncertainty-aware predictions across multiple market regimes.

A fully self-driving golf cart built from scratch — it sees the road, detects obstacles, and navigates on its own without any human input.

Led the AI and navigation subsystem of a fully autonomous self-driving golf cart built from the ground up. Developed deep learning models for real-time drivable area segmentation, obstacle detection, and sensor fusion — integrating computer vision with onboard navigation hardware for end-to-end autonomy.

A computer vision system that accurately distinguishes shadows from real objects in outdoor images — a critical capability for robotics, mapping, and autonomous navigation.

Developed a robust illumination-aware image segmentation system for outdoor environments using color space conversion, adaptive thresholding, texture entropy analysis, and morphological operations — achieving reliable shadow/non-shadow classification under variable lighting conditions.

A brain-computer interface that detects what movement a person is imagining just by reading their brainwaves — enabling hands-free control for assistive technology.

Built a DSP and ML pipeline for classifying neural activity patterns from EEG signals associated with motor imagery tasks. Applied signal preprocessing, feature extraction, and a KNN classifier to distinguish between cognitive states with high accuracy.

A fully autonomous commerce backend that handles customer support, inventory sync, and order fulfilment across WhatsApp, Messenger, and Instagram — with zero human intervention.

Engineered an autonomous LangGraph-based AI agent powered by GPT and a custom RAG pipeline (pgvector) to handle multi-channel customer support at scale. Built automated Celery Beat pipelines that sync live inventory, pricing, and stock from the Shopify Admin GraphQL API — continuously re-embedding the knowledge base so the agent's responses stay accurate in real time.

Architected the full production backend as a Docker Compose-orchestrated system: FastAPI, PostgreSQL 16 + pgvector, Redis, MinIO, and Caddy, with Alembic-managed schema migrations, HMAC-validated webhooks, and a Next.js 14 admin analytics dashboard. Integrated the Bosta shipping API for end-to-end fulfilment automation — autonomous delivery creation, AWB label generation, PDF storage, and real-time shipment tracking — replacing entirely manual logistics workflows for a live Egyptian cookware brand.

A fully offline-capable web app for managing community services, attendance, and analytics — works seamlessly with or without an internet connection across multiple user roles.

Architected a full-stack offline-first PWA serving multi-role users (servants, parents, congregants) with real-time sync via Firebase RTDB, Edge-runtime API routes, and an IndexedDB-backed conflict-resolution queue for seamless offline mutations. Features role-based access control, structured attendance tracking, aggregated analytics dashboards, Zod validation, Sentry monitoring, i18n (Arabic/English), rate-limiting via Upstash Redis, and automated CI/CD — deployed on Cloudflare Edge with sub-second response times.

A data intelligence platform that automatically collects and analyses thousands of job listings to surface trends in Egypt's labour market — all accessible through a clean desktop interface.

Engineered a full end-to-end data intelligence platform including automated web scraping, structured database management, and a polished desktop GUI. Built to aggregate, normalize, and surface labor market insights from Egypt's leading job portal.

An AI model built from the ground up that reads text and accurately determines whether the sentiment is positive or negative — without relying on any pre-existing language models.

Designed and trained a custom transformer encoder from scratch — embedding layers, multi-head self-attention, feedforward blocks — for large-scale sentiment classification. Achieved competitive accuracy on the IMDB benchmark without relying on pre-trained language model weights.

A system that automatically discovers the main themes hidden across thousands of documents — turning unstructured text into organised, searchable knowledge.

Built an unsupervised topic modelling system using Latent Dirichlet Allocation to extract and surface dominant themes across large document corpora. Designed for research and content analytics workflows requiring interpretable, scalable semantic clustering.

A tool that analyses large volumes of tweets to understand public opinion at scale — helping brands and researchers track sentiment in real time.

Developed and benchmarked a suite of NLP classifiers — LinearSVC, Naïve Bayes, Decision Trees, and MLP — for large-scale social media sentiment analysis. Produced a rigorous comparative evaluation framework for production model selection.

Original research exploring how the size and shape of an AI model affects its performance — findings that help engineers build smarter, more efficient neural networks.

Conducted an original empirical study on the interplay between neural network depth and width across diverse application domains. Findings were presented at the EURECA 2025 conference — contributing to the research community's understanding of architecture design tradeoffs.

An AI system that recognises human emotions from facial expressions in images — with built-in explainability so you can understand exactly why it made each prediction.

Designed and benchmarked a comprehensive computer vision pipeline for emotion classification, evaluating seven architectures including VGG16/19, ResNet50/101, InceptionResNetV2, Xception, and a custom CNN — with full SHAP explainability analysis.

An AI trained entirely on Shakespeare's works that can generate new text in his style — a deep exploration of how machines learn language, rhythm, and creative writing.

Trained character-level generative language models on Shakespeare's complete works using LSTM, GRU, and Bidirectional RNN architectures — exploring sequence modelling, creative text generation, and the qualitative differences between recurrent topologies.