Skip to content

04Resume

Omar Soliman

Download PDF ↓

AI & Software Engineer · London, Ontario

01Summary

Computer science student at Western University, currently an AI engineer intern at Gestalt Communications. I am strongest on the plumbing behind AI systems: ingestion and job queues that survive rate limits and restarts, document and embedding pipelines, and the Terraform and CI around them. Looking for a Summer 2027 software engineering internship where correctness matters more than demo speed.

02Experience

  1. 01Jul 2026 → Present

    AI Engineer Intern

    Gestalt CommunicationsOakville, Ontario

    • Built a pipeline ingesting United Nations procurement notices and awards to surface bidding opportunities relevant to Gestalt's services, backfilling 280k+ records on a rate limited PostgreSQL job queue.
    • Extended it with a document pipeline: attachment retrieval, PDF/DOCX extraction, Document AI OCR fallback for scanned files, sentence aware chunking, and Vertex AI embeddings stored in pgvector.
    • Persisted document text and chunks in transactional checkpoints before embedding, so failed embedding stages resume without repeating extraction or chunking.
    • Provisioned Terraform managed GCP infrastructure spanning Cloud Run Jobs, Cloud SQL, a versioned GCS archive, workload scoped IAM, Secret Manager and monitoring alerts.
    • Configured CI with PostgreSQL integration tests covering concurrent job claims, expired lease recovery and duplicate free record versioning, alongside lint and type checks.
    • Python
    • PostgreSQL
    • pgvector
    • GCP
    • Terraform
  2. 02Sep 2025 → May 2026

    Software Engineer Intern

    Swish SolarWaterloo, Ontario

    • Sole engineer on a Python pipeline converting solar SCADA and weather exports into monthly performance and data health PDFs, built to replace manual reporting.
    • Built Isolation Forest anomaly detection on interval performance ratio, power residual and AC/DC efficiency features rather than raw power output, to cut false positives from normal sunrise and sunset dips.
    • Added rule based fault categories, severity and recommended actions so operators know what to act on, and grouped consecutive zero output daylight intervals into events for reporting.
    • Python
    • pandas
    • scikit-learn

03Skills

Languages

  • Python
  • SQL
  • Java
  • C++
  • JavaScript
  • Bash

Frameworks

  • FastAPI
  • React
  • Pydantic
  • asyncio
  • Playwright
  • pandas
  • NumPy
  • seaborn

AI & ML

  • Claude API
  • Vertex AI Embeddings
  • Google Document AI
  • scikit-learn
  • XGBoost

Cloud & Tools

  • GCP
  • PostgreSQL
  • MySQL
  • pgvector
  • Supabase
  • Terraform
  • Docker
  • Git
  • GitHub Actions
  • Linux

04Education

  • B.Sc. Computer Science

    2024 to 2028 (expected)

    Western University, London, Ontario

    Dean's Honour List.