M. Hesham

About Me

heshambuilds@gmail.com 512-844-2968 LinkedIn GitHub

Hello! I am a Software Engineer with an M.S. in Computer Science and 5 years of experience building scalable backend systems, data validation platforms, and AI-powered pipelines for distributed environments. I build systems that work reliably at scale.

At Orange Business, I designed FastAPI microservices and event-driven architectures serving enterprise clients across EMEA. More recently, I engineered LLM extraction pipelines in Python that parsed unstructured data into validated, production-ready outputs, and built monitoring tools that cut system failure response times by 40%.

Location: Austin, TX

Experience

Graduate Assistant – Software Engineer

Texas State University, San Marcos, TX | Jan 2024 – Dec 2025
- Created a Python and Flask internal portal serving 70+ department users that let students quickly schedule tutoring sessions, send inquiries online, and access detailed real-time schedules, reducing staff response time to student requests by 30%.
- Developed a Python pipeline automating semester reports for a 15-person department, removing 10+ hours of manual work.
Software Engineer Intern

J&M Jewelry, San Marcos, TX | Jan 2025 – Apr 2025
- Improved order accuracy by 25% for 500+ monthly customers by engineering an LLM-powered extraction pipeline in Python that parsed natural language chat into validated ring design parameters, enforcing deterministic output schemas.
- Reduced invalid configuration outputs by 30% by implementing 25+ validation rules in Python enforcing inventory constraints.
Software Engineer

Orange Business, Cairo, Egypt | Dec 2019 – Dec 2023
- Engineered a real-time Python ingestion pipeline that aggregated distributed performance metrics across 50+ enterprise environments, cutting mean time to detection by 35% and improving alert fidelity and incident response speed.
- Reduced undetected network outages by 40% by engineering automated health-monitoring scripts in Python that continuously polled 200+ routers and switches across enterprise networks, triggering real-time alerts on anomaly detection.
- Eliminated manual compliance review for 3 EMEA enterprise clients by architecting FastAPI microservices that automated end-to-end data validation workflows, cutting processing time by 70% and enabling same-day audit readiness.

Projects

Weather Agent App

Full-Stack AI App | FastAPI, LangChain, Gemini 2.0 Flash, OpenWeatherMap API | MCP Architecture
- Built a full-stack AI-powered weather assistant using a 4-layer architecture: OpenWeatherMap API, FastAPI MCP server, LangChain agent backend, and an HTML/CSS/JavaScript frontend.
- Implemented a FastAPI wrapper around the OpenWeatherMap API to serve as an MCP (Model Context Protocol) server, exposing clean tool endpoints for the agent to call.
- Integrated LangChain with Google Gemini 2.0 Flash as the LLM to power an agentic backend capable of tool calling and multi-step reasoning over natural language weather queries.
- Built a responsive chat-style frontend in vanilla HTML, CSS, and JavaScript with message handling and error states; managed isolated Python virtual environments per service and secure API key handling via .env files.
RAG Document Assistant (Local LLM Pipeline)

AI/ML Pipeline | Python, FastAPI, LLaMA 3.1, ChromaDB, LangChain | Local Inference
- Enabled natural language querying over private documents by engineering an end-to-end RAG pipeline using Python, FastAPI, LLaMA 3.1, and ChromaDB, delivering source-cited answers across 76+ indexed document chunks.
- Eliminated cloud API dependencies by serving LLaMA 3.1 8B locally via Ollama with 4-bit quantization, enabling fully offline inference on consumer hardware with zero cost per query.
- Reduced hallucinated LLM outputs by constraining generation to semantically retrieved context using 384-dimensional vector embeddings, ensuring grounded responses tied to source documents.
- Automated document ingestion by building a modular pipeline using LangChain that recursively loads, chunks, and embeds PDFs and text files into a persistent vector store with configurable chunk sizing and overlap.
High-Performance Computing & Performance Optimization

Distributed Systems | Python, Dask, OpenMP, MPI | Multi-Node GPU Clusters (TACC Lonestar6)
- Built distributed data-processing pipelines using Python and Dask across multi-node GPU clusters, designing for scale and fault tolerance consistent with HPC and AI infrastructure requirements.
- Profiled kernel execution metrics using performance analysis tooling to optimize scheduling strategies and runtime throughput, reducing bottlenecks across parallel workloads on multi-node compute environments.
- Implemented compiler-aware optimization workflows with structured performance reporting, enabling systematic platform characterization and benchmarking across iterations.
ShopScanner for Visually Impaired

Mobile App | MediaPipe, TensorFlow | React | Accessibility
- Architected a real-time computer vision system using MediaPipe and TensorFlow to identify grocery products via mobile camera feeds.
- Implemented backend caching and model optimizations, reducing repeated lookups and improving system response time by 40%.
- Designed and deployed REST APIs bridging CV outputs with React, ensuring low-latency feedback for accessibility.
- Improved product detection reliability by 35% by engineering adaptive video processing logic in Python that handled variable lighting and environmental noise across 5+ real-world grocery store environments.
Scuba Diving Planner

Full-Stack Web App | React, FastAPI | AI & API Integration
- Reduced dive planning time by 50% building a React and FastAPI app that delivers AI-personalized dive windows across 50+ locations.
- Boosted recommendation accuracy by 40% integrating OpenWeather, Stormglass, and OpenAI APIs to aggregate wave height, current speed, and visibility into ranked dive windows.
- Eliminated 100% of invalid planning requests by designing and enforcing 20+ REST API validation rules in FastAPI, ensuring consistent and reliable data flow across the React frontend.

About Me

Experience

Projects

Skills & Certifications

Resume

Contact