Hello, I'm Mohamed Hamed.
Software Engineer.

About

Projects

Whispr Notes — On-Device Real-Time Speech Transcription

Expo / React Native app that transcribes speech in real time entirely on-device with OpenAI Whisper Base (EN) through ExecuTorch — no server, no API keys, no audio leaving the phone. 100 ms audio buffers are streamed into the model as you speak, an FSMN voice-activity model commits text after ~300 ms of silence, and raw PCM is written straight to a WAV file on disk so recording length is bounded by free space rather than memory. Transcripts are saved as searchable, taggable notes with playback and txt/json/srt export.

TypeScriptReact NativeExpoExecuTorchWhisperOn-Device MLVoice Activity Detectionexpo-router

GitHub

SNCF DelayFlow — TGV Delay Analytics Pipeline

End-to-end ETL pipeline processing TGV punctuality data (SNCF Open Data) across 3 sources (~18,000 records) with PySpark. Includes Spark SQL transformation, Parquet storage, a Random Forest Regressor (Scikit-learn) for delay prediction, and an interactive Streamlit/Plotly dashboard diagnosing root causes across infrastructure, traffic, and rolling stock dimensions.

PythonPySparkSpark SQLScikit-learnParquetStreamlitPlotlySQLite

GitHub

DataHub PFAS — Environmental Data Warehouse & BI

ETL pipeline over 929,452 measurements (CNRS Data Hub): filtering, unit conversion, PubChem API enrichment. Star schema in PostgreSQL 15 + PostGIS with 4 materialised views and 8 quality checks. MongoDB 2dsphere (~36,000 GeoJSON polygons), Flask API, Leaflet.js map and Looker Studio connector.

PythonPostgreSQLPostGISMongoDBFlaskLeaflet.jsLooker StudioPubChem API

GitHub

Skills

• Java
• Python
• TypeScript
• JavaScript
• SQL
• Bash
• Spring Boot
• Spring Security
• JPA / Hibernate
• Node.js
• REST API
• Flask
• Angular
• RxJS
• React
• Next.js
• React Native
• SCSS
• Apache Spark
• PySpark
• Apache Airflow
• dbt
• Scikit-learn
• PostgreSQL
• PostGIS
• MongoDB
• MariaDB
• SQLite
• JUnit
• Jest
• Playwright
• SonarQube
• Streamlit
• Plotly
• Looker Studio
• Tableau
• Docker
• Kubernetes
• Terraform
• Maven
• GitLab CI/CD
• GitHub Actions
• AWS S3

Work Experience

Education

Open Source Contributions

• Checkstyle

  7 merged PRs across the static analysis engine and its build: false positive fixes in IndentationCheck for new operator expressions (#18614) and an expected line separator surfaced in NewlineAtEndOfFileCheck violation messages (#18619); MultiFileRegexpHeader wired into sun_checks.xml to cover the previously uncovered Sun style rule 3.1.1 (#20812); boolean-to-enum refactors in RequireThisCheck and LineWrappingHandler; and two build fixes.

• PMD

  3 merged PRs fixing false positives in static analysis rules — two in Java (UnusedLocalVariable for pattern variables in braceless for-each, and a parser failure on switch expressions inside super() constructor calls) and one in Apex (FieldDeclarationsShouldBeAtStart with properties). Released in PMD 7.20.0 and 7.21.0.

• chalk

  Merged performance optimisation in the terminal styling library: two-argument calls such as chalk.red('hello', 'world') now use string concatenation instead of Array#join, a benchmarked 2.7x speedup on that path with no regression for other call shapes (PR #670).

• WeasyPrint

  Made the CSS text-transform property language-aware in the HTML/CSS-to-PDF rendering engine: uppercasing and lowercasing now honour the element's lang, fixing the Turkish i/İ and ı/I mappings and accented Greek characters that Python's default str.upper() and str.lower() get wrong (PR #2590, merged in v68.0).