- Contributed to the Center of Excellence’s AI strategy, driving the company’s digital transformation.
- Led end-to-end AI projects with agile methods, including RAG applications and database AI agents, plus services for machine translation, sentiment analysis, and voice transcription.
- Designed and co-managed cloud infrastructure for AI applications, optimizing for scalability and cost.
- Directed AI development processes to meet industry standards and ISO certification requirements.
Hossam Gamal Mostafa
AI Engineer
Profile Summary
AI Engineer and NLP Data Scientist with hands-on experience in system design and cloud deployment. Architects scalable, reliable AI applications, leads technical discussions with clients, and delivers production-ready LLM solutions that balance performance, cost, and maintainability.
Experience
- Delivered weekly Connect sessions for the DECI program covering Python, web development, data analysis, AI, cybersecurity, digital arts, electric circuits, embedded systems, and computer architecture.
- Mentored students and tracked their progress during sessions and office hours.
- Achieved a 4.6 average mentor rating and an 82% average graduation rate across the last two cohorts.
- Built an AI forecasting solution predicting stock quantities and costs for BarCloud customers, pairing a zero-shot forecaster fine-tuned on in-house data (new customers) with Facebook Prophet (existing customers).
- Exposed forecasts through the forecasting dashboard and the BarCloud chat interface via LLM tool calls.
- Designed a text-to-SQL solution that turns users’ questions into correct, safe MS SQL queries and returns summarized results.
- Enhanced RAG solutions for BarCloud feature questions, adding LLM serving (LlamaCpp, vLLM, Ollama) and intent classification that routes voice commands to the right pages and sections.
- Researched and implemented NLP solutions in information retrieval (multilingual NER, relation extraction) and social sciences (sentiment and aspect-based sentiment analysis, propaganda detection, offensive language identification), improving F1 and precision by 20%+ over strong baselines.
- Led discussions with technical and non-technical clients on integrating AI into their businesses, winning over 3 clients.
- Developed training plans for incoming and current junior AI developers.
- Conducted 6 intern interviews, assessing technical fit and alignment with company values.
- Represented the company at Techne Summit Cairo 2024.
British University in Egypt (BUE)
Oct 2021 – Sep 2022
Teaching Assistant of Mathematics
Full-time • On-site • Cairo, Egypt
- Delivered tutorials in Probability, Statistics, Numerical Methods, Differential Equations, and Calculus.
- Supported professors with course logistics: attendance, exam marking, and course archiving on the university’s cloud system.
- Achieved a 4.63/5 average student satisfaction score.
Projects
Product Reviews Analysis – Aspect-Based Sentiment Application
Nov 2026
Personal project • FastAPI, React, TypeScript, llama.cpp, SQLite, Docker, AWS EC2
- Built a full-stack application that scores product reviews from 1 to 5 on each aspect customers mention, with charts and CSV export.
- Served a 2B-parameter open-weight LLM with llama.cpp using JSON-schema-constrained output, so answers are always valid and limited to the aspects the user confirmed.
- Cut scoring time per review by 40% (4.0 s to 2.4 s) by profiling LLM generation, trimming the output format, and moving to a 4-bit model; runs at about 1.2 s per review on a CPU-only AWS server.
- Designed the backend for long-running work: background jobs with live progress, SQLite sessions with migrations, rate limiting, and clean failure handling when the model server goes down.
- Deployed on AWS EC2 with Docker Compose, nginx, and Caddy (automatic HTTPS) on a custom domain, with a status page showing when the demo is online.
- Wrote 200+ automated tests (pytest, Vitest) covering API flows, failures, and UI states.
Skills
- Languages
- Python, C, SQL, HTML/CSS
- LLMs & Agents
- OpenAI, LangChain, LangGraph, LlamaIndex, CrewAI
- LLM Serving & APIs
- vLLM, LlamaCpp, Ollama, ONNX, FastAPI, REST APIs
- ML & Deep Learning
- PyTorch, PyTorch Lightning, Transformers, Sentence-Transformers, LoRA, scikit-learn, GluonTS
- MLOps & Data
- Weights & Biases, DVC, Pandas, Matplotlib, Plotly
- Databases
- PostgreSQL, MS SQL, MongoDB, Neo4j
- Vector Stores
- Qdrant, ChromaDB, Weaviate, Pinecone, FAISS
- Cloud & Tools
- AWS, Azure, GCP, Docker, Git, GitHub Actions, Jira, Postman
- Professional
- Leadership, mentorship, project management, agile workflows, problem solving, teamwork
- Spoken Languages
- Arabic (native), English (fluent)
Education
Alexandria University
Sep 2015 – Jul 2019
B.Sc. in Nuclear and Radiation Engineering
Alexandria, Egypt
- Grade: 80% (Very Good), ranked 3rd in class.
- Coursework: Advanced Calculus, Linear Algebra, Differential and Partial Differential Equations, Numerical Methods, Probability, Statistics, Programming (Fortran).
- Grade: 75% (Very Good). Coursework: Calculus, Programming (C++).
Professional Development
- Program building transferable business, leadership, and problem-solving skills: adaptability, digital transformation, communication, critical thinking, and agile mindset.
- Solved the “Hustleton City Traffic Problem” case using SMART problem statements, structured thinking, and issue prioritization.