Agents and tool protocols
Agent orchestration and MCP tool integration grounded in reviewed individual project implementations.
Project evidence
AI Systems Engineer
Ahmed Aziz Ben Aissa
I design and evaluate reliable AI systems that connect agents, tools, retrieval, and production workflows.
My work connects applied AI research with responsible-AI practice, measurable evaluation, explicit safety boundaries and operational reliability.
System capabilities
AI systems, organized by evidence
Four capability groups connect the engineering focus directly to reviewed project evidence.
Retrieval and evidence
Retrieval-augmented generation, source grounding and evidence validation for research workflows.
Project evidence
Research and evaluation
Reproducible experiments, multi-seed evaluation and responsible-AI assessment with explicit limitations.
Project evidence
Production reliability
Security controls, explicit evaluation boundaries and production-oriented AI system design.
Project evidence
Flagship evidence
Three systems, with roles and limits made explicit
Each case summary separates implementation scope, personal contribution and the evidence available for public review.
Problem and system approach
An agentic evidence system integrating MCP tools, retrieval, evidence validation, security controls and evaluation.
Implementation context
The system connects MCP tool integration with retrieval and evidence grounding for European air-quality research.
Its verified scope includes evidence validation, security controls and evaluation without an unverified numeric-results claim.
- Ahmed’s role
- Sole designer and implementer; other academic group members are not represented as technical contributors.
- Work status
- Individual work by Ahmed
Verified evidence
Agent and retrieval integration: MCP tool integration connects retrieval with evidence grounding.
Method: Implementation-scope review.
Evidence controls: Evidence validation and security controls are part of the reviewed implementation.
Method: Implementation-scope review.
Evaluation boundary: Evaluation is in scope; no quantitative performance result is claimed.
Method: Implementation-scope review; no numeric outcome is claimed.
Problem and system approach
Extended FinRL–DeepSeek with a confidence-weighted, turbulence-aware signal layer for constrained PPO trading and evaluated the system through deterministic and 10-seed stochastic backtests.
Implementation context
Individual Phase 1 extension and sole report authorship, with explicit attribution to the upstream FinRL–DeepSeek project.
Results are historical backtests, not live-trading performance or investment advice; upstream attribution and documented evaluation limitations must remain visible.
- Ahmed’s role
- Individual Phase 1 extension and sole report authorship, with explicit attribution to the upstream FinRL–DeepSeek project.
- Work status
- Individual work by Ahmed
Verified evidence
Deterministic EP90 annual return: 17.12%
Method: Deterministic EP90 historical backtest.
Ten-seed stochastic mean annual return: 16.843% ± 3.517%
Method: Ten-seed stochastic historical backtest; mean ± standard deviation.
Ten-seed stochastic mean maximum drawdown: −25.921% ± 1.671%
Method: Ten-seed stochastic historical backtest; mean ± standard deviation.
Problem and system approach
A multi-agent Socratic tutoring system with persistent learner state, learning history, adaptive guidance and user-triggered fairness-audit functionality.
Implementation context
Verified implementation includes a planner agent, Socratic tutor, learning-signal extractor, fairness reviewer, controller and state transitions, SQLite persistence, a CLI and a Streamlit interface.
Fairness audits are user-triggered; no learning-effectiveness or fairness-outcome evaluation has been completed.
- Ahmed’s role
- Individual academic project.
- Work status
- Individual work by Ahmed
Verified evidence
Multi-agent architecture: Planner agent; Socratic tutor; learning-signal extractor; fairness reviewer; controller and state transitions.
Method: Implementation-scope review.
Persistent learner state: SQLite persistence supports learner state and learning history across sessions.
Method: Implementation-scope review of the CLI and Streamlit interfaces.
Responsible-AI boundary: Fairness audits are user-triggered; learning effectiveness and fairness outcomes remain not evaluated.
Method: Implementation-scope review; no learning-effectiveness or fairness-outcome result is claimed.
Professional experience
AI engineering in enterprise and production contexts
Public descriptions stay at an approved scope and do not imply access to employer code or private systems.
Apprenticeship
Ayming
AI Engineer
October 2025–presentParis, France
Building enterprise AI systems for knowledge retrieval and workflow automation.
- Developing workflows for company enrichment and technology detection.
- Building AI tools supporting grant analysis and consultant decision-making.
Internship
VroomVroom
AI Systems Engineer
June 2025–August 2025Paris, France
Built a production-oriented AI assistant orchestrating tools and APIs for multi-step user workflows.
- Developed a modular AI agent architecture integrating 12+ tools.
- Built ingestion pipelines for thousands of driving-school records and reviews.
- Contributed to backend integration, containerization and deployment.
Employer work; implementation details and source code are not public.
Selected technical work
Publication follows evidence review
Only evidence-reviewed work is published here; additional technical case studies remain under validation.
Technical toolkit
Skills grouped by system function
Displayed skills resolve to published project or professional evidence; proficiency scores are intentionally omitted.
Agent and retrieval systems
Capabilities grounded in approved agent, retrieval and workflow evidence.
AI/ML research and evaluation
Research methods and evaluation practices tied to approved project evidence.
Production and data engineering
Production capabilities evidenced by approved professional responsibilities.
Programming foundations
Programming foundations used across the approved work record.
Certifications
Verified learning and platform credentials
Three high-signal credentials are featured first, with additional certifications across AI, data science, deep learning and cybersecurity available below.
Credential 01
Microsoft
Microsoft Certified: Azure AI Apps and Agents Developer Associate
Issued July 2026 · Expires July 2027
Credential 02
Anthropic
Introduction to Model Context Protocol
Issued March 2026
Credential 03
Hugging Face
AI Agent Course
Issued April 2025
View all certifications (8 more)
Credential 04
Microsoft
Microsoft Certified: Azure Data Scientist Associate
Issued July 2025 · Expired July 2026
Credential 05
Google
Connect and Protect: Networks and Network Security
Issued March 2025
Credential 06
Google
Play It Safe: Manage Security Risks
Issued March 2025
Credential 07
Google (Coursera)
Foundations of Cybersecurity
Issued February 2025
Credential 08
Hugging Face
AI Agents Fundamentals
Issued February 2025
Credential 09
NVIDIA
Fundamentals of Deep Learning
Issued September 2024
Credential 10
NVIDIA
Applications of AI for Predictive Maintenance
Issued July 2024
Credential 11
NVIDIA
Building Transformer-Based NLP Applications
Issued June 2024
Education
Mathematics, computer science and AI systems
Only approved institution, programme, location and date facts are shown.
2024–2026
Aivancity School for Technology, Business and Society
Programme Grande École
Paris, France
2021–2024
Paris Dauphine–PSL, Tunis campus
Bachelor’s degree in Mathematics and Computer Science for Decision Making
Tunis, Tunisia