Ahmed S.

Within Amundi, Europe's leading asset manager, I work as a Software Engineer / AI Engineer on the Alto ESG Metrics ecosystem, a central platform used by ESG analysts to manage and leverage extra-financial data integrated into responsible investment processes. Beyond developing the platform's core backend, I designed and developed a conversational artificial intelligence layer allowing analysts to query ESG data in natural language, via an architecture based on the MCP (Model Context Protocol) protocol.

Software Engineering

- Development and maintenance of the Alto ESG Metrics backend in Node.js / NestJS, exposing ESG data to analysts via high-performance, robust, and tested REST APIs

- Design of the application architecture (modules, services, DTOs, validation) ensuring the platform's scalability and maintainability

- Design and evolution of the Angular user interface, enabling ESG analysts to view, filter, and leverage companies' extra-financial metrics in their portfolio

AI Engineering

- Design and development of a conversational platform (Python, FastAPI, FastMCP) allowing ESG analysts to query ESG metrics in natural language

- Implementation of MCP tools that integrate natively with the Alto ESG Metrics API to retrieve, aggregate, and deliver requested extra-financial data in real-time

- Design of the orchestration between the conversational agent and existing backend services: tool contracting, context management, response structuring

- Establishment of a decoupled architecture between the AI layer (FastAPI/FastMCP) and the business backend (NestJS), ensuring security, traceability, and reusability of tools exposed to the agent

Design and development of an internal conversational AI platform for Athoria's sales, HR, and operational teams.

The objective was to centralize company knowledge, connect business tools, and allow employees to search for information, reason on internal data, and execute actions directly from a secure conversational interface.

The platform relies on an agent-based architecture with enterprise RAG and MCP integrations, to assist sales teams in responding to RFPs by suggesting the most relevant consultant profiles based on ERP data and business context.

- Design of the technical architecture and development of the internal conversational AI platform

- Implementation of an agent-based architecture with LangChain and LangGraph for orchestrating complex AI workflows and executing business actions

- Development of a RAG (Retrieval-Augmented Generation) pipeline connected to internal documentation, operational data, and company business tools

- Development of a custom MCP with FastMCP (Python) connected to BoondManager to assist sales teams in automatically identifying the best consultant profiles for RFPs

- Integration of Microsoft Outlook and Calendar tools for reading, writing, and executing actions with human validation (human-in-the-loop)

- Implementation of observability for AI agents and LLM pipelines with Langfuse (tracing, monitoring, analytics, debugging)

- Development of an evaluation framework for AI models and workflows using an “LLM-as-a-Judge” approach

- Deployment and management of infrastructure on AWS

- Backend development with Python / FastAPI (hexagonal architecture) and frontend development with React.js

Within the CISIRH (Interministerial Center for Human Resources IT Services), attached to the Ministry of Economy, I contributed to the development of the CTDSN application (Concentrator Translator for Nominative Social Declaration), a critical solution responsible for collecting, transforming, and transmitting nominative social declarations from public servants to recipient organizations (URSSAF, pension funds, etc.).

This mission is part of the modernization of the State's HR IT system and compliance with regulatory obligations related to the DSN.