Elwalid Aboulaakoul

About Me

Salam! I'm Elwalid Aboulaakoul, an Engineering student specializing in Information Systems and Big Data at ENSA Berrechid. I work at the intersection of MLOps, distributed systems, and AI research, with a focus on building production-grade systems and exploring sovereign AI architectures.

My interests span from agentic AI systems and RAG pipelines to performance optimization and decentralized inference. When not coding, I'm fascinated by vintage ships and historic trains, which inspire my sense of craftsmanship and engineering elegance.

Work Experience

Software Engineer Intern (MLOps & Cloud) - Intelcia

June 2025 - September 2025 & September 2024 - October 2024

• Architected end-to-end MLOps pipeline for call volume forecasting using Temporal Fusion Transformers (TFT) processing monthly records
• Deployed modular training and inference workflows on Google Cloud Platform, replacing manual processes with fully automated, serverless architecture
• Implemented "Freeze-Eval" quality gate system to automatically prevent model regression before production promotion
• Exposed predictions via high-performance FastAPI microservice with auto-scaling capabilities
• Built feature-flagged NL→SQL retrieval layer for ConverseQuery RAG app (chat-to-data), routing queries between local data and BigQuery with automatic legacy fallback; added fast-mode toggles to isolate bottlenecks
• Achieved ~24% median latency improvement and ~7.7% average speed-up on standardized KPI queries

Software Engineering Fellow - Headstarter

July 2024 - October 2024 | Remote

• Completed intensive software engineering fellowship focused on full-stack development and AI integration through weekly project builds and deployments

Data Science Fellow - ALX, African Leadership Group

2023 - 2024 | Remote

• Completed intensive ML engineering program covering end-to-end model deployment

Projects

Deep Research Agent for Arabic Scholarship

Built agentic research system processing 200M+ tokens across 434MB Arabic corpus from Qatar Digital Library, combining web search with RAG for multi-step reasoning with full tool execution visibility.

• Engineered semantic search pipeline (extraction → chunking → multilingual embeddings → ChromaDB) with metadata preservation enabling precise cross-reference retrieval across classical Arabic texts
• Deployed on Parallax decentralized GPU infrastructure via custom Model Context Protocol (MCP); OpenAI-compatible streaming gateway delivers responses in <10s with automatic failover
• Integrated web search capabilities using Serper, Jina, and Exa APIs for comprehensive research workflows
• GitHub Repository

Technical Skills

Languages: Python, TypeScript, C/C++, CUDA, Java, Matlab, Bash

Data Science & ML: PyTorch, W&B, Hugging Face (Transformers, Diffusers, Datasets), MLFlow

Full Stack: React (Native), Next.js, FastAPI, Node.js, Spring Boot, REST APIs, HTML, CSS

Databases: PostgreSQL, SQLite

DevOps & Cloud: Docker, CI/CD (GitHub Actions), Terraform, GCP

Dev Environment: NeoVim, Zed, Zsh, Zellij, Git, uv, Codex/Claude Code

Awards & Certifications

• Moroccan National Informatics Olympiad (2022) - Semi-Finalist (Top 50 Nationwide) - Ranked in top 50 nationwide, solving complex algorithmic problems
• LLM Agents Certification - UC Berkeley (November 2024)

Research Interests

Exploring distributed AI systems, sovereign intelligence architectures, and agentic workflows. Interested in contributing to research at the intersection of systems optimization and machine learning deployment.

Contact

Feel free to connect with me: