SOFTWARE ENGINEERING

Zakariye Mohamed

Backend & Infrastructure Software Engineer

BuildingReliable backends

I'm a computer science student and software engineer focused on building reliable backend systems, cloud infrastructure, and AI-powered applications. I've worked across site reliability, systems engineering, software development, and AI engineering.

Download resume PDF

Other versions

Infrastructure resume AI engineering resume

GitHub LinkedIn

zakariyem0099@gmail.com

Atlanta, GA, USA

Open to opportunities

About Me

Backend systems, infrastructure, and applied AI

At Georgia State University, I'm studying computer science while building across the stack: operations tools, job orchestration, containerized services, and ML pipelines. My work connects application development with the infrastructure that keeps it running.

Skills & Expertise

Tools I use across my work and projects

Languages

Python
C++
Java
JavaScript
TypeScript
SQL

Backend

FastAPI
Django
REST APIs
WebSockets

Cloud & Infrastructure

AWS
Azure
Docker
Kubernetes
Terraform
CI/CD

AI/ML

PyTorch
TensorFlow
RAG
LLMs
MCP

Databases & Observability

PostgreSQL
Redis
Prometheus
Grafana
ELK
CloudWatch

Work Experience

Systems, reliability, software, and AI engineering

Systems Engineer Intern

InComm Payments

September 2026 – Present

Support client migration to an SFTP batch-processing platform, documenting file flows, dependencies, and operational handoffs.
Build, test, and maintain monitoring alerts for client file workflows and incident detection.

SFTP
Monitoring
Batch processing

Site Reliability Engineer Intern

IBM

May – August 2026

Built Sentinel, a React, Django, and PostgreSQL operations platform for application health, incidents, logs, and on-call workflows.
Added correlation-ID tracing across two ELK clusters, Jira and Gemini incident automation, and an MCP knowledge layer.
Developed Terragrunt/Terraform templates and Python automation; evaluated the OpenShift Kube Descheduler Operator in a proof of concept.

Django
PostgreSQL
ELK
Terraform
OpenShift

AI Engineer Fellow

Cornell Tech

April 2026 – Present

Built and evaluated neural-network, RAG, and multimodal LLM applications with TensorFlow and Keras.
Worked with company partners to translate product requirements into AI features and MCP workflows.

TensorFlow
Keras
RAG
MCP

Software Engineer Intern

Signify

January – May 2026

Automated build, test, artifact publishing, and deployment with Python, Bash, Bamboo CI, Docker, and Artifactory.
Migrated workloads to AWS and Azure and programmed C/C++ BLE Mesh software for real-time device control and state synchronization.

Python
C++
Docker
AWS
Azure
Bamboo CI

AI Engineer Intern

Crescent Technology

July – December 2025

Built a production RAG pipeline and optimized retrieval and response latency.
Implemented expiring conversational memory and MCP workflows for lead capture, intent routing, and escalation.
Deployed containerized AI microservices on AWS with Docker and REST APIs.

Python
RAG
MCP
AWS
Docker
REST APIs

Education

Expected May 2027

Georgia State University

B.S. Computer Science

Major GPA: 3.8. Coursework includes algorithms, operating systems, systems-level programming, and machine learning.

Admitted

Georgia Institute of Technology

M.S. Computer Science

Expected enrollment after graduating from Georgia State in May 2027.

Featured Projects

Backend systems, infrastructure, and applied AI. Open a project for the engineering details.

AI Agent Orchestration · Team project

Abyss

A collaborative AI agent marketplace with queued execution, MCP tools, and spending controls.

Python
FastAPI
TypeScript
React
+2

Job queues, cancellation, and selective retries.
Configurable tools and per-session spending limits.

Implemented outcome

Selective retries retain completed task output; configurable spending limits put bounds on agent jobs.

Engineering details

Contributed MCP integrations, per-session tools, session templates, and file attachments.
Added job queues, cancellation, selective task retries, and daily, weekly, and monthly spending limits.
Improved reconnect event handling, responsive layouts, keyboard access, and result previews.

Full stack: Python, FastAPI, TypeScript, React, WebSockets, MCP

Team project / fork of jtdao24/abyss.

GitHub

Source-based flow / strings, lists, TTL

Systems Programming

Redis Clone

A Redis-like TCP server in C with RESP parsing, string/list operations, and key expiration.

C
TCP sockets
RESP
CMake

Multiplexed TCP connections with select().
String/list commands and millisecond key expiration.

Implemented outcome

TCP clients can execute string, counter, and list commands with conditional writes and expiring keys.

Engineering details

Implemented a select-based loop to accept and handle multiple TCP client connections.
Added RESP command parsing and in-memory string, counter, and list operations.
Implemented EX/PX expiration, TTL/PTTL queries, conditional SET, and key deletion.

Full stack: C, TCP sockets, RESP, CMake

GitHub

Source-based architecture / multi-node platform

Infrastructure & GPU workloads

Homelab Infrastructure

A multi-node platform for backend services, local inference, RAG, and GPU training.

FastAPI
Docker Compose
PostgreSQL
Redis
+3

Worker leases, retries, and Redis wakeups.
Cross-node monitoring and GPU-backed workflows.

Implemented outcome

One job control plane connects queued backend work, document retrieval, local inference, and GPU training.

Engineering details

Built a PostgreSQL job control plane with worker leases, retries, and Redis wake notifications.
Configured Caddy, Tailscale networking, and Prometheus/Grafana monitoring across services and GPU workloads.
Connected Ollama inference, document ingestion and retrieval, and LoRA training through backend APIs.

Full stack: FastAPI, Docker Compose, PostgreSQL, Redis, Qdrant, PyTorch, MLflow

Services GPU worker Web interface

Private source / access required.

Architecture

Source-based pipeline / training to inference

Backend & ML delivery

Housing Price MLOps

A housing-price pipeline from feature engineering to tracked models and REST inference on AWS.

Python
FastAPI
XGBoost
Optuna
+3

Optuna tuning with MLflow experiment tracking.
Container delivery through GitHub Actions to ECS.

Implemented outcome

Tracked training artifacts feed a FastAPI prediction service with a configured ECR/ECS delivery pipeline.

Engineering details

Built preprocessing, feature engineering, training, and evaluation pipelines with tracked model artifacts.
Added Optuna tuning and MLflow experiment logging for XGBoost models.
Served predictions through FastAPI and configured GitHub Actions delivery to ECR and ECS.

Full stack: Python, FastAPI, XGBoost, Optuna, MLflow, Docker, AWS ECS

GitHub

Source-based architecture / persistent finance data

Full-Stack Fintech

Kash

An expense tracker with AI receipt entry, recurring transactions, and budget alerts.

Next.js
React
Prisma
PostgreSQL
+3

Authenticated accounts and transaction workflows.
Gemini receipt extraction and scheduled budget checks.

Implemented outcome

Receipt extraction, recurring entries, and scheduled budget checks support an end-to-end expense workflow.

Engineering details

Implemented authenticated transaction and account workflows backed by Prisma and PostgreSQL.
Integrated Gemini receipt extraction and recurring-transaction processing.
Added spending analytics and scheduled budget checks with email notifications.

Full stack: Next.js, React, Prisma, PostgreSQL, Clerk, Gemini, Inngest

GitHub

Source-based model / text, video, audio

Multimodal AI

EMOTION

A multimodal emotion and sentiment pipeline combining text, video, and audio on MELD.

Python
PyTorch
BERT
OpenCV
+2

BERT, R3D-18, and audio CNN feature fusion.
Held-out evaluation and SageMaker deployment code.

Implemented outcome

A single fused model produces seven emotion classes and three sentiment classes, with held-out evaluation code.

Engineering details

Combined BERT text embeddings, an R3D-18 video encoder, and a 1D audio CNN through feature fusion.
Implemented separate seven-class emotion and three-class sentiment heads.
Evaluated train/dev/test splits with accuracy and weighted precision, with TensorBoard and CloudWatch logging.

Full stack: Python, PyTorch, BERT, OpenCV, SageMaker, TensorBoard

GitHub

More Projects (3)

Source-based flow / audio to model insights

Audio ML & model visualization

Audio CNN Visualizer

An environmental-sound classifier with an interactive interface for inspecting predictions, audio waveforms, spectrograms, and CNN feature maps.

Python
PyTorch
Torchaudio
Modal
+1

Residual CNN trained on the 50-class ESC-50 dataset.
Top-three predictions with per-layer feature visualizations.

Engineering details

Implemented a residual convolutional model with optional intermediate feature-map outputs.
Built a Modal inference endpoint that resamples audio, computes mel-spectrograms, and returns top-three class probabilities, waveforms, and layer visualizations.
Added a Next.js upload interface; training code includes MixUp, TensorBoard logging, and saved model checkpoints.

Full stack: Python, PyTorch, Torchaudio, Modal, Next.js

GitHub Live demo

Collaborative full-stack project

QuickBite

Customer and driver workflows with chat-based ordering, location-aware restaurant discovery, and real-time order messaging.

Next.js
TypeScript
Supabase
OpenAI

Structured AI actions connect conversations to cart and order workflows.
Driver tools include delivery updates and an AI message copilot.

Engineering details

Structured AI actions connect conversations to cart and order workflows.
Driver tools include delivery updates and an AI message copilot.

Full stack: Next.js, TypeScript, Supabase, OpenAI

Collaborative project; repository forked from bobbleteeb/Team-Sleep.

GitHub Live demo (sign-in)

Cloud delivery

Amazon Prime CI/CD

Delivery configuration for a React streaming-interface clone, with Jenkins pipelines, container packaging, and Kubernetes deployment files.

Jenkins
Docker
Terraform
AWS ECR/EKS
+2

Configured quality analysis, filesystem scanning, Docker builds, and ECR publishing stages.
Added Terraform infrastructure and Kubernetes deployment configurations.

Engineering details

Configured quality analysis, filesystem scanning, Docker builds, and ECR publishing stages.
Added Terraform infrastructure and Kubernetes deployment configurations.

Full stack: Jenkins, Docker, Terraform, AWS ECR/EKS, SonarQube, Trivy

GitHub

Open Source & Collaboration

MCP tools & session workflows
Integrated an MCP tool catalog with configurable per-session tool access, reusable templates, and file attachments.
Backend job execution
Added job queues, cancellation, reruns, and selective retries that retain completed task output.
AI spending controls
Implemented per-job budgets and daily, weekly, and monthly spending limits.
Real-time reliability
Improved WebSocket reconnect handling, event deduplication, stopped-job states, and result previews.
Responsive & accessible interfaces
Refined responsive layouts, keyboard access, scene lifecycle handling, and reduced-motion support.

Certifications

Professional credentials in cloud engineering and machine learning

eCornell, Cornell University

Machine Learning Foundations

Verify credential

Microsoft

Microsoft Azure Fundamentals

Issued July 2025

Credential ID: F79DD55FE9602050

Amazon Web Services

AWS Certified Cloud Practitioner

Issued June 2025

Credential ID: dcda45f9248e4d0e8a586a36373ebebb

Get In Touch

Have a project or role in mind? Get in touch.

Contact Information

Email

zakariyem0099@gmail.com

Phone

+1 (470) 962-8351

Location

Atlanta, GA, USA

Follow Me

GitHub LinkedIn

Send a Message

Send a message directly through this form.

Playing as Player

A little side quest

Wanna play a game?

Jump over the glitches. Keep your AI Twin running.

Score 0000Best 0000

Space / Up to jump - Tap on mobile

High scores all players

Finish a run to set the first high score.

Zakariye Mohamed

About Me

Skills & Expertise

Languages

Backend

Cloud & Infrastructure

AI/ML

Databases & Observability

Work Experience

Systems Engineer Intern

Site Reliability Engineer Intern

AI Engineer Fellow

Software Engineer Intern

AI Engineer Intern

Education

Georgia State University

Georgia Institute of Technology

Featured Projects

Abyss

Redis Clone

Homelab Infrastructure

Housing Price MLOps

Kash

EMOTION

Audio CNN Visualizer

QuickBite

Amazon Prime CI/CD

Open Source & Collaboration

MCP tools & session workflows

Backend job execution

AI spending controls

Real-time reliability

Responsive & accessible interfaces

Certifications

Machine Learning Foundations

Microsoft Azure Fundamentals

AWS Certified Cloud Practitioner

Get In Touch

Contact Information

Email

Phone

Location

Follow Me

Send a Message

Wanna play a game?

High scores all players