AboutExperienceWorkContact
Open to Opportunities

Ahmed Saad

6x Hackathon Winner — Turning Complex Problems into Deployed Solutions. Specializing in AI Agents, Web3, and Distributed Systems.

View Portfolio
GitHub
Ahmed Saad
About Me

Crafting Digital
Experiences

I'm a Software & Networks Engineering student at INSAT who builds with an ownership mindset — not just writing code, but taking products from 0 to deployment. At Sofrecom I shipped a Web3 CSR platform that boosted transparency and user engagement.

A 6x Hackathon Winner across national and international competitions, ranked top 0.11% globally in IEEEXtreme, and Vice President of ACM INSAT managing a 500+ student tech community.

10+
Projects
6x
Hackathon Wins
2+
Years Exp

Experience

Career Timeline

Sofrecom

Full Stack Engineer Intern

Jul – Aug 2024

Built a production-grade Web3 CSR platform at Orange's subsidiary in Tunisia, boosting corporate transparency by 40% via blockchain integration. Designed and consumed RESTful APIs (Node.js/Express), implemented gamification logic that increased user engagement by 30%, and owned the full lifecycle from MongoDB schema design to Angular frontend delivery.

Node.jsExpressMongoDBAngularBlockchain

INSAT

Software Engineering Student

2023 - 2028

Tunisia's leading engineering school. Focusing on distributed systems, algorithms, and AI architecture. Top 0.11% globally in IEEEXtreme.

Vice President, ACM INSAT Chapter — leading technical workshops and events for a 500+ student community.

Selected Work

Award Winning Projects

FAIN — AI Trading Ecosystem
🥉 3rd Place — IHEC CodeLab 2.0
Read Case Study →

FAIN — AI Trading Ecosystem

Institutional-grade trading terminal for the BVMT — 12 microservices, on-the-fly XGBoost price forecasting (RSI/MACD/Bollinger features), PPO Reinforcement Learning portfolio optimization with SHAP explainability and real macro data (World Bank/IMF/BCT), Gemini+Perplexity sentiment pipeline, and a Llama 3.3 70B RAG chatbot for BVMT/CMF docs.

FastAPI •NestJS •Next.js •XGBoost •ChromaDB •Docker
Moufida — Local-First AI Ecosystem
🏆 AI Minds Hackathon — 24h Sprint
Read Case Study →

Moufida — Local-First AI Ecosystem

Privacy-first, offline-capable AI desktop copilot built in a 24h hackathon sprint. 5 Python microservices behind a Tauri + Next.js overlay: multimodal ingestion (CLIP/BLIP/Whisper), explainable vector search via a custom XQdrant fork, DBSCAN file organisation, Piper TTS + faster-whisper STT copilot, and knowledge-gap detection — all running on a Qwen3 4B model over local network (ngrok).

Tauri •Next.js •FastAPI •XQdrant •MongoDB •Qwen3
QDesign — Explainable AI for Biological Discovery
🏆 1st Place — Vectors in Orbit
Read Case Study →

QDesign — Explainable AI for Biological Discovery

🏆 1st Place — Vectors in Orbit (Qdrant × InstaDeep × YellowSys). Co-architected a multimodal biological research workbench: forked Qdrant into XQdrant (Rust) to decompose ESM-2 cosine similarity into biological properties (α-helix, surface charge), plus a custom SciAgents-inspired HITL Co-Scientist (Planner/Scientist/Ontologist/Critic agents via OpenRouter) that drafts IEEE-formatted papers.

Rust •NestJS •Next.js •XQdrant •OpenRouter •ESM-2
VeriVest — Real Estate Tokenization
🏆 Best Pitch — AIxCYBER
Read Case Study →

VeriVest — Real Estate Tokenization

🏆 Best Pitch — AIxCYBER. Tunisia's first AI-driven, bank-backed, blockchain-based digital asset wealth platform. Built on Hedera HTS/HCS for fractional property ownership (1000 tokens per asset) with n8n AI workflows for automated price prediction and portfolio recommendation.

Next.js •Hedera •Express.js •n8n •MongoDB
AstroVerse — AI Business Strategy Platform
🏆 1st Place — AI for Entrepreneurs
Read Case Study →

AstroVerse — AI Business Strategy Platform

🏆 1st Place out of 30+ teams in a 22h sprint. A 4-service platform (Next.js 15.5 + 2× FastAPI + n8n) where 11 specialized AI agents generate investor-ready business documents — Incubator, Market Analyst, Strategist, Brand Architect, Product Manager, Legal Assistant, CMO, CRO, Production Ops, Growth Strategist, and Business Planner — each rendered as professional PDFs via WeasyPrint + Jinja2.

Next.js 15 •FastAPI •n8n •WeasyPrint •SQLModel
Agent IA BH Assurance — Le Conseiller Augmenté
🥇 1st Place — NEXT Challenge (Solo)
Read Case Study →

Agent IA BH Assurance — Le Conseiller Augmenté

🥇 1st Place — NEXT Challenge by BH Assurance (Solo vs teams of 3, month-long Aug–Sep). A 4-service multi-agent RAG platform: FastAPI + Qdrant (Qwen3-Embedding-8B, 4096-dim) + FlashRank re-ranker + Qwen3-4B via Ollama. Solo-engineered the entire data pipeline — extracting, translating, and structuring BH Assurance's product catalog from multilingual unstructured PDFs/CSVs using Gemini 2.5 Pro.

FastAPI •Qdrant •Ollama •Qwen3 •FlashRank •n8n
JCR Platform — Serverless AI Course Recommender
🥇 1st Place — 4h Sprint
Read Case Study →

JCR Platform — Serverless AI Course Recommender

🥇 1st Place — 4h JCR Sprint (April 2025). Built and deployed a full-stack serverless ML platform in 4 hours: trained an XGBoost + Random Forest ensemble for multi-label course recommendation, deployed on Cloudflare Workers with Neon Serverless Postgres, Drizzle ORM, Hono.js, and bcrypt auth — all edge-optimized with sub-50ms global response times.

XGBoost •Random Forest •Cloudflare Workers •Neon Postgres •Drizzle ORM
INSAT Student App — My First Real Project
🎓 First Serious Project
Read Case Study →

INSAT Student App — My First Real Project

The project that taught me how software actually works. A full-stack desktop application for INSAT students: a 5947-line PyQt5 frontend (grade tracking, INSAT orientation simulation for GL/RT/IIA/IMI, spider/line/bar charts, AI reports via Llama 3.1 8B, 3 languages) backed by a 2258-line Cloudflare Worker API (50+ handlers, Neon Postgres, bcryptjs auth, Cloudinary, email notifications).

PyQt5 •Cloudflare Workers •Neon Postgres •Llama 3.1