Available for freelance

Engineering
Full-Stack & AI
Systems That Ship

I build high-performance full-stack web applications and custom AI systems that optimize business operations and protect revenue.

View ProjectsHire Me
About me

Bridging the Gap
Between Code & AI

I build production software — full-stack applications, AI-powered systems, and data pipelines that solve real business problems. My work spans everything from fraud detection engines to distributed booking platforms, shipped for real users, not just class assignments.

I don't just integrate AI — I architect it. From predictive risk models to RAG chatbots to retail analytics, I design complete products that turn raw data into decisions clients can act on.

Robust Architecture

Scalable systems designed for high-throughput workloads and low-latency delivery.

Model Integration

Seamlessly connecting LLMs and custom ML models to production-grade APIs.

Data Engineering

Building the pipelines that feed intelligence, from raw streams to vector stores.

Tech Stack

What I work with

Languages & Frontend
PythonJavaJavaScriptTypeScriptReact.jsNext.jsTailwind CSSHTML5CSS3

Comfortable across the stack — from typed frontend interfaces to backend services and scripting.

Backend & APIs
NestJSNode.jsExpress.jsFastAPIDjangoREST APIsMicroservicesRBACAuthentication Systems

Designed and built backend services and APIs, from monolithic to distributed microservice architectures.

AI & Machine Learning
Decision TreesRandom ForestXGBoostIsolation ForestANNCNNRNNLSTMBERTTransformers

Experience building and evaluating models — from classical ML to deep learning and transformer architectures.

NLP, LLM & Vision
LangChainRAG PipelinesFAISSChatbot DevelopmentYOLOv8Computer VisionTensorFlowPyTorchKeras

Built production RAG pipelines and computer vision models, backed by the core Python ML/DL toolchain.

Data & Databases
PostgreSQLMySQLMongoDBRedisNeo4jPrismaEDAFeature EngineeringApache KafkaPySparkETL Pipelines

Worked with structured and unstructured data — from relational and graph databases to full ETL pipelines.

DevOps & Workflow
DockerGitGitHubRabbitMQn8nPostmanJestJiraAgile / ScrumCI/CD

Comfortable shipping with modern tooling — containerized services, queues, automation, and testing practices.

Work

Selected Projects

AI & Risk Analytics

AI-Powered Industrial Insurance Fraud Detection

Problem: Insurance companies lose millions to fraudulent industrial claims that are hard to detect manually.
Approach: Layered Isolation Forests, XGBoost, and neural networks to isolate anomalies inside financial profiles — cross-referencing document analysis, image forensics, and behavioral pattern detection.
Outcome: Automated fraud flagging pipeline that catches suspicious claims before payout, reducing manual review load across the claims team.
Read Case Study →
AI & Automation

RAG Chatbot

Problem: Companies can't give employees instant, accurate answers from internal documents.
Approach: Built a multilingual AI chatbot (EN/FR/AR) trained on company PDFs, with source attribution and an auto-retraining feedback loop.
Outcome: Deployed in production, giving employees instant, source-cited answers instead of manual document searches.
Read Case Study →
RAG Chatbot1 / 2
NLP & Fine-tuning

BERT Sentiment

Problem: Businesses struggle to understand customer sentiment at scale across thousands of reviews.
Approach: Fine-tuned a BERT model on labeled review data and served it via a REST API for real-time inference.
Outcome: 90.98% classification accuracy, enabling real-time sentiment monitoring at scale.
Read Case Study →
BERT Sentiment
Big Data & Pipelines

Scientific Knowledge Graph

Problem: Researchers waste hours manually searching thousands of scientific papers to find collaborators, topic trends, and related work.
Approach: Built a full Big Data pipeline collecting 2,300+ papers from arXiv and CrossRef, mapped into an interactive knowledge graph with a 6-page Streamlit dashboard.
Outcome: Surfaces collaboration networks, topic trends, and paper recommendations via TF-IDF similarity and Louvain community detection.
Read Case Study →
Data Engineering & ML

Stockout Risk Predictor

Problem: Retail teams find out products are out of stock only after sales are already lost, and treat every point-of-sale the same despite very different performance patterns across locations.
Approach: Feature engineering pipelines and exploratory data setups built on real retail data to trace localized product depletion risk, paired with K-Means clustering to segment points-of-sale by behavior.
Outcome: A KPI dashboard flagging at-risk stores before stockouts happen, feeding directly into store-level restocking decisions.
Completed — source confidential (client data)
Stockout Risk Predictor1 / 8
Distributed Systems

TaxiBook

Problem: Booking systems crash under load and can't scale across multiple locations.
Approach: Built a real-time ride booking platform distributed across 4 machines with automatic failover between nodes.
Outcome: Zero-downtime booking system that stays available even when individual machines fail.
Read Case Study →
TaxiBook1 / 2
Microservices

Stock Management

Problem: Inventory systems go down when one service fails, blocking the entire operation.
Approach: Designed an event-driven inventory platform where each service runs and fails independently.
Outcome: If one service fails, the rest of the platform keeps working — no single point of failure.
Read Case Study →
Stock Management1 / 2
B2B & Microservices

MedSupply Manager

Problem: Healthcare supply chains have no secure way to manage who can order what — pharmacies, hospitals, and suppliers all need different access levels.
Approach: Built the authentication backbone of a B2B medical supply platform — JWT auth, role-based permissions for 5 user types, and the admin dashboard, with a CI/CD pipeline gated by SonarQube.
Outcome: Secure, role-based ordering flow across 5 distinct user types, deployed with automated quality gates on every release.
Read Case Study →
MedSupply Manager
Healthcare

MediKeep

Problem: Clinics manage patient records manually, leading to errors, lost data, and slow appointments.
Approach: Built a full-stack healthcare platform with role-based access for doctors and patients, appointment scheduling, and an analytics dashboard.
Outcome: Replaced manual record-keeping with a structured system that reduces errors and speeds up appointment handling.
Read Case Study →
MediKeep1 / 5
Experience

Where I've worked

Jun 2025 – Jul 2025
Data Science Immersion Program
DSIP4 · Djezzy × MajestEYE · Algiers,Dar El Beida, Algeria

Intensive 6-week data science immersion program. Built real-world retail analytics projects: out-of-stock prediction and point-of-sale segmentation. Covered the full data product lifecycle — data governance , data quality, EDA, feature engineering, ML model development, KPI dashboards, and product pitch.

PythonPandasScikit-learnEDAMajestEYE Tools
May 2025 – Jul 2025
AI & Backend Developer Intern
Expert Petroleum Services · Boumerdes, Algeria

Built and deployed a multilingual RAG chatbot (EN/FR/AR) using LangChain, FAISS, and local LLMs. Developed RESTful APIs with Node.js, JWT auth, and RBAC. Implemented auto-retraining feedback loop and full test suite.

LangChainFAISSNode.jsMongoDBJWTREST APIs
Dec 2024 – Jan 2025
Software Engineering Intern
Sonatrach · Boumerdes, Algeria

Studied enterprise DBMS architecture, network infrastructure in one of Algeria's largest industrial environments.

DBMSNetworkingSoftware Architecture
Services

Full-Stack + AI

Full-Stack Web Architectures

End-to-end web applications—from structured database schemas and robust backend APIs to modern, interactive user interfaces.

  • Scalable backend architecture
  • Responsive, production-ready UI
  • Secure database design

Intelligent Automation & Workflows

Eliminating slow, manual business tasks by engineering reliable automation pipelines, third-party API configurations, and custom document tools.

  • Automated business workflows
  • Third-party API integrations
  • Custom document & data tools

Predictive Risk & Intelligence Engines

Custom machine learning and deep learning infrastructure designed to parse business data, discover anomalies, and prevent financial loss.

  • Anomaly & fraud detection
  • Predictive forecasting models
  • Data-backed decision systems
Contact

Let's build
something

Let's discuss your project parameters. Book a strategy brief or submit your development constraints directly below.

Connect

Feel free to reach out!

Download CV

Usually reply within 24 hours — no bots, just me.