Skip to content
View AhmedShehata2002's full-sized avatar
🧨
Working
🧨
Working

Block or report AhmedShehata2002

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
AhmedShehata2002/README.md

AHMED SHEHATA


I INVESTIGATE COMPLEX SYSTEMS.

BIOLOGICAL · FINANCIAL · BUILT


Data & Strategy | Health & Environmental Analytics | Co-founder @ AEROME | MSc Health Data Science | CFA Level I Candidate
Abu Dhabi, UAE · English & Arabic




01 / SELECTED WORK


BEYOND TISSUE OF ORIGIN

If we read a cancer's genes, can we predict how well a drug will work — better than just knowing which organ it came from?

End-to-end machine-learning study of 8 anti-cancer drugs across ~700 cancer cell lines, joining GDSC2 drug-response data (Wellcome Sanger) with DepMap genomics (Broad Institute). A 20% test set is locked away before any training, all tuning and gene selection happen inside cross-validation folds, and every model has to beat an organ-of-origin baseline, not just the average. Genes beat that baseline for 7 of 8 drugs (r = 0.42–0.76, bootstrap 95% intervals), and SHAP recovered the textbook biomarkers — BRAF, TP53, BCL2 — as the top inputs for their drugs without being told.

Python scikit-learn LightGBM SHAP Pandas

→ View Repository


PREVISIT

Can a bilingual pre-visit history replace the paper intake form in UAE private clinics — and does the evidence actually hold up?

A working synthetic-data demo (v0.1) of a bilingual Arabic/English pre-visit intake tool: one shared questionnaire engine behind both a WhatsApp-style simulator and a secure patient form, layered and versioned consent, deterministic safety rules that run ahead of any AI step, AES-256-GCM encrypted storage, and an append-only audit trail. No real patient data.

Paired with a full evidence, law and economics dossier in the repo README — grounded in peer-reviewed documentation-burden and pre-visit-planning research, an honest read of a comparable real-world deployment's actual completion rates, and UAE health-data law (PDPL, Federal Law 2/2019, DOH/DHA standards, MOHAP medical-software rules) — every claim marked Verified, Hypothesis, or Illustrative rather than asserted.

Claude API FastAPI Next.js PostgreSQL Python TypeScript

→ View Repository


PASSIVEROI

Which early passive-design choices cut a building's cooling electricity — and do they actually pay back?

Streamlit app that turns 2025 NASA POWER climate data for Abu Dhabi, Dubai and Cairo into a transparent, component-based sensible-cooling model (conduction, glazing solar gains transposed to each facade with pvlib, air exchange, internal gains), then converts the estimated savings into payback and discounted NPV. Negative NPVs stay visible and evidence gaps are labelled rather than hidden. Working local v0.1 with an automated pytest suite and a validation notebook.

Python Streamlit pvlib Plotly pytest

→ View Repository


AEROME CASE STUDIES

What does environmental data look like when it has to win a commercial pitch?

Two delivered projects from AEROME, the environmental design studio I co-founded. My part is the data layer around the architecture team's design work: DesignBuilder simulation output, site climate data, UAE building-code and market benchmarks, and cost data, combined into the business case. Zawaya Mall (Egypt): −42% cooling cost and −35% total energy. Dubai villa: 67.9 kWh/m²/yr, 53% below the Al Sa'fat code baseline.

DesignBuilder Autodesk Forma Energy Benchmarking Commercial Analysis

→ View Repository


More work — credit-risk ML, SQL churn analysis, health-data AI apps — in all repositories ↗




02 / RESEARCH


HOW BUILDINGS AFFECT HUMAN HEALTH

Evidence from Air Quality, Stress, and Epigenetics

Narrative Review · Preprint · 2026

Examines environmental exposure, the built environment, and human health — spanning air quality, stress physiology, and epigenetic mechanisms.

DOI: 10.5281/zenodo.22012930 ↗




03 / EXPERIENCE


AEROME · Co-Founder
Data, Strategy & Environmental Analytics
Cairo, Egypt · Mar 2026 — Present

Python and DesignBuilder-based workflows analysing climate, energy, thermal comfort, and ventilation before major design and capital decisions. Projects modelling up to 42% lower cooling demand and 35% lower overall energy demand. Commercial analysis and market development with real-estate developers.


M42 · Data Science Intern
Environmental Health Analytics
Abu Dhabi · G42 & Mubadala · Jul — Oct 2025

Extracted, cleaned, and validated 50K+ laboratory records from LIMS data. Identified geographic and seasonal patterns to inform monitoring priorities. Presented findings at the 2nd Water Quality Forum to scientists, government officials, and senior stakeholders. Built an LLM-enabled research tool with conversational literature querying.


MASTERPIECE ENGINEERING SERVICES · Data Analyst
Abu Dhabi · Apr 2024 — Apr 2025

Analysed and validated 50K+ records using SQL and Excel, reducing reporting inconsistencies by 25%. Built 10+ interactive dashboards. Produced quantitative insights across healthcare technology, hospitality, and construction for commercial proposals.




04 / EDUCATION


UNIVERSITY OF BIRMINGHAM
MSc Health Data Science
Jan 2026 — Sep 2027 · Dubai, UAE

CFA CANDIDATE
CFA Program · Level I · Nov 2026
Ethics · Quantitative Methods · Economics · Financial Reporting & Analysis
Equity · Fixed Income · Derivatives · Portfolio Management

UNIVERSITY OF TEXAS AT AUSTIN — McCOMBS
PGCert Data Science & Business Analytics
2024 — 2025 · Abu Dhabi, UAE

OXFORD BROOKES UNIVERSITY
BSc (Hons) Medical Science · 2:1
Sep 2020 — May 2023 · Oxford, UK




05 / TOOLS


DATA & PROGRAMMING   Python · Pandas · NumPy · Scikit-learn · LightGBM · SHAP · SciPy · TensorFlow / Keras · SQL

ANALYTICS & AI   Statistical Analysis · Machine Learning · EDA · NLP · Generative AI

BUSINESS INTELLIGENCE   Excel · Power BI · Tableau · Streamlit

FINANCE & STRATEGY   Financial Statement Analysis · Quantitative Methods · Market Analysis · Commercial Analysis




06 / ELSEWHERE


LinkedIn ↗   Website ↗



Abu Dhabi, UAE · Open to roles in health data science, investment analysis, and data-driven strategy

Pinned Loading

  1. Personal_Projects Personal_Projects Public archive

    Python

  2. Portfolio- Portfolio- Public archive

    HTML