Sandra Binu

Data Science and AI professional with hands-on experience in RAG, Machine Learning, and Computer Vision

Talent

London AreaMember since May 13, 2026

RAGBig DataMachine LearningTableauComputer VisionAzure

Hire this person

Send a job offer directly to this candidate

RAGBig DataMachine LearningTableauComputer Vision

About

I am a Data Science and AI professional with hands-on experience in RAG, Machine Learning, and Computer Vision. I am currently working as an Innovation Engineer at Alten UK, contributing to a team-based GraphRAG solution for the aerospace industry using Azure-based tooling and collaborative DevOps workflows. With strong technical skills in developing predictive models and implementing data-driven solutions, I am committed to delivering impactful, scalable solutions that improve processes, generate insights, and support innovation.

Experience

Innovation Engineer at Alten Innovation Lab (2025-09 – Present)

Developed an aerospace-domain chatbot for information retrieval and contextual question answering using Retrieval-Augmented Generation (RAG).
Built and optimized retrieval pipelines using Microsoft Graph RAG, Light RAG, and Haystack, improving performance on advanced mathematical and technical benchmark questions to nearly 90% accuracy through effective prompt engineering.
Integrated RAG-Anything to support retrieval of equations and complex technical content, while contributing to data workflows and deployment practices using Git, virtual machines, Azure DevOps, and Azure Blob Storage.

Artificial Intelligence and Machine Learning Intern at IMMCO Inc. Saranga Geosoftware and Engineering Services Pvt. Ltd. (2022-06 – 2022-12)

Developed an Automatic File Updating Tool, which reduced employee work time for the latest documents by 80%, and designed an Automatic Attendance Register Tool using Face Detection that helped reduce manual register time by 20% during company events.
Implemented data augmentation and image processing pipelines to improve product defect classification accuracy by 5%, and leveraged Faster R-CNN and YOLOv5 for custom object detection, achieving over 90% precision on test datasets.