● Published
PhD candidate · Lead AI Engineer · University Assitant Lecturer
Researching how agents improve themselves. Building the systems that run them.
AbdElKader Seif El Islem Rahmani — PhD research on generative-AI methods for autonomous agents and self-improvement behavior, alongside production AI engineering (LLM workflows, RAG, agent systems, backend infrastructure) and university teaching.
Google Scholar · ResearchGate · LinkedIn · GitHub · ORCID
6Published papers
9Teaching assignments, 2025–2027
70+Production releases, Systelium
9Universities teaching at
Autonomous, self-improving agents
Focus. Generative-AI methods for autonomous agents and self-improvement behavior, PhD in Computer Science (Artificial Intelligence), University of Naama.
Areas. Autonomous and self-improving agents, LLM systems, adaptive multi-agent systems, reliable AI architectures.
Intelligent agents
→ Retention / memory Voice assistant
→ Self-improvement Stochastic analysis · Survey
→ Adaptation Non-stationary settings
→ Multi-agent collaboration MIND-META
Research evolution
- Sep 2026 · SN Computer Science
Architectural analysis: a systematic survey of generative and cognitive self-improving agents. - Sep 2026 · NCIIT26
Adaptation: monitoring-based strategy for non-stationary multi-agent collaboration, plus two co-authored Arabic NLP reviews. - Apr 2026 · NCMAI'26
Mechanism: isolating stochastic self-improvement in LLM agents. - Oct 2025 · SPECOM25 Proceedings
Retention and stateful interaction: a lightweight voice-assistant architecture with privacy-preserving design.
Published work
Published papers, book chapters and conference contributions. DOIs are linked where assigned.
● Published
MIND-META: A Monitoring-Based Adaptive Strategy for Non-Stationary Multi-Agent Collaboration
multi-agentadaptationnon-stationarity● Published
Semantic Annotation of Arabic Text: A Pipeline-Centric Review
Arabic NLPsemantic annotationreview● Published
Arabic Dialect Identification on Social Media: A Review of Datasets, Models, and Emerging Trends
Arabic NLPdialect identificationreview● Published
Characterizing Stochastic Self-Improvement in Autonomous LLM Agents: Mechanism Isolation and Boundary Condition Analysis
self-improvementLLM agents● Published
Retention-Augmented Voice Assistant: A Lightweight Architecture for Stateful Interaction with Comprehensive Evaluation and Privacy-Preserving Design
memoryvoice assistantsprivacyUniversity teaching
Courses, tutorials (TD) and practical work (TP) at the universities of Naama and Saida.
2026–2027
1st semester
Computer Architectures
Applications of Deep Learning in Finance
Introduction to AI (practical work)
2026-2027
1st semester
Human Machine Interface
Computer Architectures
2025–2026
2nd semester
Computer Tools
Language Theory
Mathematical Logic
1st semester
Data Analysis
What gets built and shipped
Systelium
Problem. Enterprise workflows with heavy manual document handling.
System. AI platform on FastAPI, PostgreSQL, async processing, microservices, Docker, CI/CD, with Google Gemini in production workflows.
Result. 70+ production releases, 99.9% uptime, 70% less manual effort.
AgriTechly
System. Plant disease detection models with end-to-end ML workflow: data prep, training, evaluation, optimization, deployment.
Result. +15% model accuracy, −20% inference latency.
Independent consulting
Scope. 20+ client projects end-to-end. AI-focused since January 2023: LLM applications, RAG pipelines and agent systems.
Prodigy InfoTech
Problem. Completed a selective ML internship program, applying Python and NLP techniques to solve real-world prediction and classification tasks.
System. Improved data preprocessing pipelines and trained baseline models with 10%+ performance gains.
Result. Documented results and shared learning outcomes, aligning with industry best practices for transparency and reproducibility.
Capabilities
- Agents / AI
- Agentic AI, LLMs, RAG, embeddings, multi-agent systems, evaluation, tool use, structured outputs
- Backend
- Python, FastAPI, Node.js, TypeScript, JavaScript, PostgreSQL
- Infrastructure
- Docker, CI/CD, GitHub Actions
- ML tooling
- PyTorch, TensorFlow, Hugging Face, LangChain, LangSmith, W&B
- Governance
- ISO/IEC 42001 Lead Implementer training, AI risk and lifecycle management
Research ↔ engineering
Research asks what agents can become. Engineering finds out what survives contact with production.
ResearchArchitectureImplementationExperimentDeploymentImprovement
Curriculum vitae
AbdElKader Seif El Islem Rahmani
Experience
- Lead AI Engineer (Freelance), Systelium — 06/2025–present
- Deep Learning Engineer, AgriTechly — 07/2023–06/2024
- Independent Freelance Developer / AI Consultant — 01/2020–present
Education
- Ph.D. Computer Science | Artificial Intelligence, University of Naama
- M.Sc. Computer Science | Computer Modeling of Knowledge and Reasoning, University of Saida
- B.Sc. Computer Science | Information Systems and Software Engineering, University of Saida
Certifications & achievements
- ISO/IEC 42001 Lead Implementer Certificate, PECB
- Huawei Seeds for the Future 2024
- 2× Hackathon 2nd place
Why intelligent agents
I research how LLM-based agents can retain, reflect and adapt, and I build AI systems for real clients. Each side corrects the other: production shows where agent ideas break, and research gives structure to what I build.
Currently: PhD work on generative-AI methods for autonomous agents, AI platform engineering as a technical owner, and teaching computer science and AI courses at the universities of Naama and Saida. Languages: Arabic (native), English (fluent), French (conversational).
Get in touch
Start a WhatsApp conversation
LLM / RAG applicationAI agent systemBackend & deploymentDeep learning / visionAI consultingAI governanceResearch collaboration
Profiles
Google Scholar
Scholarly record and citations.
ResearchGate
Research presence and papers.
Career record.
GitHub
Code and repositories.
ORCID
Record of all indexed papers