Scarecrow: ML for Aircraft Engine Management¶

Project done at Deloitte (Client: Rolls-Royce)

Technologies used: Python, streaming machine learning, Data bricks Timeline : Feb 2021 - Jan-2023

Publications: Presented Scarecrow - Intelligent Annotation platform for Engine Health Management in AI ML Systems conference

White papers: Demonstrating online learning on Rolls-Royce blogs

Impact: Preventive maintenance identified with 15% less false positivity(estimated)

Team: 7-10 member team consisting of data engineers and data scientists

Problem Statement: Assisting subject-matter experts (SME) in identifying various performance issues in an engine

1. Built and presented the POC of "Human with AI" tool called Scarecrow in a hackathon in London and got appriciated by CEO and CTO of R2Factory (subsidiary of Rolls Royce).
2. Led the team in designing, building and deploying the "Human with AI" tool in various use cases. Led the team in building a full scale web based framework that continuously learns (streaming machine learning) by observing the decisions taken by SMEs.
3. Built failure identification and prediction models for two parts of a type of aircraft engine and implemented it on more than 1000 engines. This was done by monitoring SME's who look at data from different sensors from aircraft engines in flight to identify the engines\parts that may need maintenance or have low performance. These models are used to predict failures among engines and provide a list of engines for the SME's to focus on and assists in saving 1200 man hours.