Vol. 26 / No. 1 Louisville, KY · Alexandria, EG Open August 2026

Software engineer · new-grad SDE

Mohamed Konsowa

Mohamed Konsowa
Available
Aug ’26

I build the boring, necessary half of AI: APIs, Terraform, monitoring, and the services that keep agents honest in production.

Dual degree · GPA 3.94 · ECPC finalist 2022 & 2023

01

Field log

I treat AI as infrastructure. A model is only useful if it sits behind real APIs, a gateway, a database, and a dashboard someone can page at 2am.

This year I work with GE Appliances through the HIVE AI Innovation Center at UofL: enterprise agent workflows on one side, Cloud Run, MCP auth, and Terraform monitoring on the other. Before that I built VR and multimodal systems at the Center for Human Systems Engineering, mentored NSF REU students, and turned a research codebase into microservices.

2+2 dual degree: B.S. Computer Science & Engineering at the University of Louisville, B.E. Artificial Intelligence Engineering at Alamein International University. GPA 3.94 at both. Dean’s List at UofL. Top of class at AIU for three years. Arabic native, English fluent.

02

Work index

  1. 01 GE Appliances now
  2. 02 HIVE AI now
  3. 03 CHSE ’24–’25

Software Engineering Intern, Enterprise Collaboration

GE Appliances, via HIVE AI Innovation Center

  • Partnered with GE engineers on an enterprise agentic AI platform: sprint planning, pull requests, spikes, and architecture reviews.
  • Evaluated MuleSoft Flex Gateway and API governance for MCP servers, including policies, health probes, and authorization with OPA.
  • Designed agent orchestration with n8n, REST, and MCP so agents can call tools and talk to GE systems.
  • Proposed a two-horizon platform roadmap: autonomous maintenance and an employee workflow builder with rule-based authority checks.

MuleSoft · n8n · MCP · OPA · Agile

Software Engineering Intern, Platform and Infrastructure

HIVE AI Innovation Center · University of Louisville

  • Contributed to a microservice agentic AI platform on GCP: Cloud Run, Cloud SQL/PostgreSQL, REST APIs, fault-tolerant service design.
  • Implemented JWT authentication for the MCP template: endpoint validation, token verification, production hardening.
  • Built Terraform monitoring across DEV, QA, and PROD: uptime, 4xx/5xx, error logs, SQL health, latency dashboards that cut time-to-diagnose.

GCP · Terraform · Cloud Run · PostgreSQL · JWT · MCP

Software Engineering Intern

UofL Center for Human Systems Engineering · full-time summer, part-time year

  • Built a VR diagnostic environment in Unreal Engine with eye-tracking and live ML inference for early dementia detection. BERT, CNNs, DNNs at 90%+ accuracy.
  • Shipped Flask/FastAPI backends, refactored legacy Python into microservices (−30% runtime errors), and wired audio and gaze pipelines in Unity and Unreal.
  • Improved model recall 15% with Optuna and ensembles. Collected HCI data from 50+ participants. Mentored 5 NSF REU students. Co-authored a manuscript on AI-driven VR diagnostics.

Unreal · Unity · PyTorch · Optuna · FastAPI · HCI

03

Selected builds

  1. 01

    Agentic Web App Generator

    GE Appliances prototype. Multi-agent planning, code generation, and sandboxed execution.

  2. 02

    RagyVerse

    RAG document Q&A. FastAPI, React, LangChain, FAISS. Ingestion, chunking, embeddings, retrieval.

  3. 03

    MCP Tool Integration Service

    MCP server for calendar, filesystem, and APIs. Shared schemas for plug-and-play LLM tools.

  4. 04

    Scylla

    VAE-LSTM and VAE-CNN anomaly detection on logs and metrics. Dockerized Flask, PostgreSQL, alerts.

  5. 05

    VR Dementia Assessment

    Unreal medical room, Vive Pro gaze tracking, live CNN/BERT scoring over backend APIs.

  6. 06

    AMD Detection

    Capstone with AIU, UofL, Deusto. Fine-tuned ResNet50. Precision +15%. Deployable inference.

  7. 07

    Driver Gaze Prediction

    Scene-aware gaze targets and a scoring loop for novice-driver attention alignment.

04

The kit

Languages

C++, Python, Java, JavaScript, TypeScript, SQL, C, HTML/CSS

Core CS

Data structures & algorithms, OOD, distributed systems, operating systems

Backend

FastAPI, Flask, Django, Node, Express, REST, microservices, JWT, OAuth, MCP, n8n

Cloud

AWS, GCP, Cloud Run, Cloud SQL, Terraform, Docker, GitHub Actions, CI/CD

Data

PostgreSQL, MongoDB, MySQL, FAISS, vector DBs, Firebase, SQLite

AI / ML

PyTorch, TensorFlow, Hugging Face, scikit-learn, LangChain, RAG, Optuna, OpenCV

Ops & security

GCP Monitoring, Cloud Logging, alert policies, API governance, MuleSoft Flex Gateway, OPA

Also

React, Unreal Engine, Unity, Linux, Arduino, Git, Agile/Scrum

How a platform is wired

Click a layer.

Pick a layer to read the note.

05

Margin notes

School

University of Louisville
B.S. Computer Science & Engineering
Expected Aug 2026 · GPA 3.94 · Dean’s List ×2

Alamein International University
B.E. Artificial Intelligence Engineering
Oct 2021 – Aug 2026 · GPA 3.94 · Top of class ×3

Paper

NVIDIA Building AI Agents with Multimodal Models · Apr 2026

ITI Cyber Security, 120 hrs · Aug–Sep 2023

ITI Embedded Systems, 156 hrs · Aug–Oct 2022

Ink

ECPC Finalist 2022 & 2023. Led a team in Egypt’s largest collegiate programming contest.

2nd, AI Hackathon PDF Q&A with Hugging Face Transformers.

Dean’s List / Top of class UofL twice. AIU three years.

AI Olympiad judge Vision, NLP, embedded systems, software design.

06

Colophon

Write

mkonma647@gmail.com (click to copy)

Download resume (PDF)

github.com/MohamedAKonsowa

LinkedIn

Visual notes

Warm paper, hairline rules, numbered rows instead of a card grid: Abdul Kader Safi and Grafik.

Press G for the layout grid: Corentin Bernadou.

Oversized name as architecture, portrait sitting on the type: Danielle Levitt.

Rotated spine on the margin: Shelby Kay.

Type

Newsreader for the broadsheet. IBM Plex Mono for the running head, indices, and the stamp.

© 2026 Mohamed Konsowa. One accent. No cards. No glow.