Open to opportunities

Kanupriya Pandey

@kanupriyapandey

Message

I build production data and ML pipelines across cloud platforms.

United Kingdom

Message

What I'm looking for

I'm looking to build end-to-end production data and ML systems, from large-scale ingestion and processing through model integration and cloud deployment, while continuing to learn new problem domains.

I'm building and deploying production data pipelines on Azure with PySpark, alongside ML pipelines and LLM-driven agentic systems integrating models with vector databases at Lumanity.

At Cell and Gene Therapy Catapult, I built and operated AWS cloud and DevOps workflows using Docker, CodeBuild, S3, EFS, Flask, Dash, and Streamlit. I developed an automated peak-detection algorithm that saved more than 300 hours of manual signal analysis and ran reliably in a live data pipeline.

I've built data-integration pipelines for heterogeneous, high-dimensional datasets and trained classification models in PyTorch and TensorFlow, including graph neural networks and Random Forest models. I also delivered a real-world evidence ML model that improved on a legacy model's accuracy while offering greater control and transparency.

Earlier, I analysed large genomic datasets on HPC at Barts Cancer Institute and built Python and SQL reporting workflows at Mphasis. I enjoy taking systems from prototype through ingestion, processing, model integration, and deployment.

Experience

Work history, roles, and key accomplishments

Current

Data Scientist

Current

Lumanity

Jan 2026 - Present (8 months)

Building and deploying production data pipelines on Azure cloud with PySpark, and ML pipelines in production. Engineering LLM-driven, agentic pipelines that integrate models with vector databases.

Azure PySpark ML Pipelines LLM Vector Databases Docker CI CD

Data Scientist

Cell and Gene Therapy Catapult

Jan 2022 - Jan 2025 (3 years)

Built and operated cloud and DevOps workflows on AWS, including Docker for application deployment, CodeBuild for CI/CD, and S3/EFS for data storage. Developed and productionised a signal-processing algorithm for automated peak detection, saving 300+ hours of manual analysis.

AWS Docker CodeBuild CI CD S3 EFS Flask Dash Streamlit PyTorch TensorFlow graph neural networks Random Forest

Research Analyst

Barts Cancer Institute

Feb 2021 - Aug 2021 (6 months)

Built a custom Python data-processing pipeline on HPC to analyse large genomic datasets at scale. Applied PCA, k-means clustering and rigorous statistical testing to derive signal from noisy, high-dimensional data.

Python HPC PCA ANOVA KS Test FDR Genomics