C#/.NET services behind financial consulting workflows
LyesHadjar.
I build backend systems, applied AI products, and Arabic language technology that hold up beyond the demo.
3 cards. Use the left and right arrow keys to move through them.
Systems with
a reason to exist.
Retrieval, language, and secure-product work presented as engineering decisions—not a wall of technology badges.
Algerian Legal GraphRAG Assistant
A bilingual French/Arabic legal assistant combining vector search, BM25, and knowledge-graph traversal over Algerian law.
QUERY -> VECTOR + BM25 + GRAPH -> GROUNDED ANSWERArabic Government RAG System
A document-grounded Arabic retrieval system over an estimated 1,200+ official government documents.
ARABIC QUERY -> NORMALIZE + RETRIEVE -> SOURCE PASSAGESEncrypted Chat Application
A cross-platform encrypted messenger using AES-256 payload encryption and RSA key exchange with no plaintext persisted at rest or in transit.
MESSAGE -> AES-256 -> RSA KEY EXCHANGE -> PEERShipped inside
real constraints.
Government, industrial, automotive, and financial work across overlapping engineering roles since 2024.
Oct 2025 - PresentAI / Software Engineer InternCommand Strategy AdvisoryParis, France - Remote
Client-facing backend tools and financial consulting workflows built for production delivery.
- Translate requirements from financial consultants into production-deployable C#/.NET REST API solutions across an estimated 3+ active client engagements.
- Automate reporting workflows with reusable backend modules, reducing turnaround by an estimated 40%.
Jun 2025 - Nov 2025AI / Software Engineer InternMinistry of Higher Education & Scientific ResearchAlgiers, Algeria - Hybrid
An Arabic question-answering system grounded in official government documents.
- Built an end-to-end Arabic QA system with AraBERT, CAMeL Tools, and FAISS over an estimated 1,200+ official documents.
- Designed the FastAPI backend and React interface, reaching an estimated sub-two-second average response time across 5+ government data sources.
Jul 2025 - Sep 2025AI / Software Engineer InternHolding Algeria Chemical Specialities (ACS - ENAP)Bouira, Algeria - Hybrid
A content platform and retrieval assistant designed for independent business-team operation.
- Delivered a Next.js 14 and TypeScript company site with SSR and SEO optimization, reaching an estimated Lighthouse score above 90.
- Integrated Payload CMS and built a LangChain/ChromaDB assistant indexing an estimated 500+ documents.
Sep 2025 - Oct 2025AI / Software Engineer InternMinistry of Vocational Training and EducationAlgiers, Algeria - Hybrid
A national orientation engine matching learner profiles to training and career pathways.
- Built a Rust-backed NLP recommendation engine matching learner profiles against an estimated 200+ training programs.
- Added explainability and career-path mapping across an estimated 5+ professional domains.
Jul 2025 - Nov 2025Software Engineer & Design Team LeadK2A AutoRemote - Contract
Automotive SaaS product engineering paired with a reusable visual system.
- Architected and shipped AI-powered features for an automotive SaaS product in Next.js.
- Established a Figma system spanning an estimated 40+ components to improve design-to-development handoff.
Jun 2024 - Aug 2024Artificial Intelligence InternThynk Tech DZAlgiers, Algeria
A classification pipeline integrated into an existing business-critical application.
- Developed and deployed a Python/scikit-learn classification pipeline processing an estimated 10,000+ records.
- Reduced manual review time by an estimated 30% within a six-week sprint.
Arabic language.
Embedded vision.
Two research tracks, both built around difficult real-world constraints.
Algerian Dialect Diacritization - Dataset & Modeling
Research Team Leader · ENSIA
Co-leading the first Algerian dialect diacritization corpus: 6,086 sentences with 95.2% character-level inter-annotator agreement, built through a three-phase human annotation pipeline over LLM pre-annotation.
Preparing a paper that benchmarks BiLSTM-CRF, AraBERT, MARBERT, and DziriBERT under MSA/dialect transfer-learning strategies.
Optimization of Vision Transformer Models for Embedded Object Detection
Selected Final-Year Research Project · ENSIA × Supervised by Sid-Ahmed Berrani
Investigating optimization strategies for ViT-based object detectors on resource-constrained embedded platforms.
Targeting lower inference latency, memory use, computational cost, and energy consumption while preserving detection performance.
Engineer.
Researcher.
Community builder.
Final-year B.Sc. Artificial Intelligence Engineering student at ENSIA, expected to graduate in June 2027.
Vice President of the ENSIA Tech Community, after serving as AI Cell Manager and Secretary General. Native in Arabic and Kabyle; professional English and intermediate French.
Supporting club-wide strategy, operations, and coordination across technical and community teams.
Languages
Python · C# / .NET · Rust · TypeScript / JavaScript · Dart / Flutter
Backend & delivery
FastAPI · .NET · REST APIs · PostgreSQL · MongoDB · Docker · Git · GitHub Actions · CI/CD
AI & retrieval
LangChain · LlamaIndex · GraphRAG · Arabic NLP · FAISS · ChromaDB · Qdrant · Hugging Face · NetworkX
Machine learning
scikit-learn · TensorFlow · Sentence Transformers · AraBERT · MARBERT · DziriBERT
Product
React · Next.js · Flutter · Figma · Adobe Illustrator · Prototyping
Bring me the
hard problem.
Open to backend and AI engineering roles, technical collaborations, and research work.
lyes.hadjar.04@gmail.com