Championed the definition of development processes, tools, and infrastructure for producing and serving AI models
Championed the definition of the AI products’ operating model
Analyzed and developed solutions for multi-cloud deployment using Ray Serve, Langchain, and Kubernetes on AWS, Azure, and GCP
Designed a production environment to deploy PyTorch models with GenAI and Agentic AI applications, replicable across AWS, Azure, and GCP using Ray Serve
Built two iterations of an engineer copilot AI system for aircraft problem diagnosis and downtime estimation
Defined taxonomy for aircraft and engine malfunctions and built a temporal knowledge graph consumed by an LLM agent
Built a graph database using a fine-tuned LLM and Langchain to classify aircraft maintenance records
Fine-tuned a PyTorch BERT classifier on 20M maintenance records, deployed on Hugging Face for AWS, Azure, and GCP
Defined a RAG architecture and LLM agent workflows using Langchain’s LangGraph library
Used Flask and Streamlit for quick prototyping
Used FastAPI to interface agents with SAP ERP web services for spare part availability and procurement reporting
Implemented vector databases: Azure AI Search, Pinecone, and Vertex AI Matching Engine
Used ChatGPT-4o on Azure/AWS and Gemini on GCP
Created a two-stage GenAI/Agentic AI RAG architecture to propose causes, parts for inspection/replacement, expected downtime, and subsequent parts worth checking
Built multimodal GenAI POCs accepting image or voice input and generating prompts for the engineer copilot application
Used Microsoft Copilot on Azure for chatbot orchestration and tool/agent calling flow definition
Defined data architecture and migration to create an analytics DWH using Snowflake, consolidating data from multiple airlines
Built a data lake on AWS, Azure, and GCP for different OPCOs
Defined data governance and data quality roles with HR
Created data pipelines for ingestion
An Enterprise architect with Data scientist and Artificial intelligence background, approaching 15 years of experience, gained across UK, the EU, Africa, Middle east, Asia Pacific and Australia regions. The core competence covers Enterprise architectural definition, Data science, Artificial intelligence with the use of LLM for Generative AI, creation of neural networks for the creation of classification models or for video or text classification, Time series analysis, all developed using Python.
Usually required to provide (Team & solution) leadership and know how, mastering the inter relationship between the 4 architectures a complex organization relies upon (Business, Data, Infrastructure, Application) to deliver its business strategy, liaising, depending upon the situation, with business or technical stakeholders, domain owners, project steering committees.
Discover other experts with similar qualifications and experience
2025 © FRATCH.IO GmbH. All rights reserved.