Kasper Heyst Thomsen - Senior Data Analyst

Journey

Scroll through the stack to reveal detailed impact

Senior Data Engineer

PAYHAWK

BigQuery ETL Python

2025 - Present

100+

Table
Dependencies

Expanding BigQuery databases to support global spend management.
Managing complex dependencies across 100+ core tables for data integrity.
Architecting automated pipelines to provide real-time executive dashboarding.

Scan Program Analyst

HP

Redshift LookML Modeling

2023 - 2025

Scale

Global Data
Models

Architected enterprise-level data models in Redshift for scan program tracking.
Developed high-impact Looker & PowerBI dashboards for internal stakeholders.
Analyzed complex user behavior datasets to drive global program optimizations.

Data Analyst

HP

SQL AI Integration Analysis

2022 - 2023

AI

Internal
Innovation

Boosted Instant Ink adoption rates by identifying and removing friction points.
Spearheaded internal AI training and knowledge sharing for the analytics team.
Delivered SQL-based churn prediction insights to drive customer retention.

Analytics Lead

VITEC

Tableau Strategy BI

2020 - 2022

Lead

Technical
Roadmaps

Defined the long-term technical analytics roadmap for core product suites.
Managed high-value client consulting for custom BI & data tracking solutions.
Optimized development pipelines to ensure faster delivery of actionable reports.

Analytics Consultant

VITEC

R SQL Automation KPI

2018 - 2020

400+

Automated
Workflows

Engineered automated R and SQL workflows for over 400+ clients.
Built real-time internal KPI dashboards to monitor operational performance.
Streamlined reporting cycle times by 60% through custom data automation.

Technical Arsenal

Data Engineering

BigQuery Snowflake dbt Redshift PostgreSQL Docker Airflow Jenkins

Analytics & BI

Python SQL Looker LookML Tableau Power BI R Plotly

Cloud & ML

GCP AWS Scikit-learn TensorFlow Ollama AI Integration Azure LLMs

Education

MSc Business Intelligence

Aarhus University | 2017 – 2019

Specialization in Forecasting, Text Analytics, and Data Mining.

BSc Economics & Business Admin

NTU Singapore | 2016 – 2016

Focus on:

• International Finance
• Communications Strategies

BSc Business Administration

Aarhus University | 2014 – 2017

Thesis on consumer behavior in autonomous vehicles (Grade: A).

Key Projects

Click a project to explore the architecture

Applied AI

Smart Home Automation & Local LLM

Built an ecosystem of 130+ automated devices via Home Assistant. Integrated a locally hosted Large Language Model (Ollama) for intuitive natural language voice control.

Raspberry Pi Ollama YAML

This ongoing project integrates Ollama with Home Assistant's voice pipeline, allowing conversational control over 130+ privacy-focused devices.

Key Features:

• 130+ Devices automated
• Local, private LLM logic
• Custom YAML & Python scripting

Smart Home Automation & Local LLM

Raspberry Pi Ollama YAML Python

This ongoing personal project is a deep dive into the world of home automation and applied AI. Starting with a Raspberry Pi and Home Assistant, I've built a system that now controls and automates over 130+ devices throughout my home.

The current focus is on enhancing user interaction through natural language. I am integrating and optimizing a locally hosted Large Language Model (LLM) using Ollama with Home Assistant's "Assist" voice control pipeline. This allows for more intuitive and conversational control of smart devices.

Key Features & Technologies:

Platform: Raspberry Pi, Home Assistant
Device Count: 130+ (lights, sensors, media, appliances, etc.)
AI Integration: Local LLM (Ollama), Home Assistant Assist
Focus: Privacy (local processing), responsiveness, natural language

Enterprise

Global Scan Data Architecture (HP)

Architected robust data models using Looker & LookML for massive global datasets. Enabled deep behavior analytics and stakeholder dashboards for HP's scan ecosystem.

Looker LookML SQL

Built massive global behavior models for HP scan devices, providing critical insights for product teams via PowerBI.

Objectives:

• Robust LookML Modeling
• User behavior & friction tracking
• Executive reporting dashboards

Game Dev

Custom Video Game & AI Playground

Developed a 2D game using Phaser.js and Vite. Explored practical application of LLMs (ChatGPT, Suno, Gemini) for AI-driven generation of game assets and music.

Phaser.js AI Assets

Built from scratch to learn game physics loops while applying advanced generative AI models for music and sprites.

Learnings:

• Game Physics & Animation
• Suno/ChatGPT Asset Gen
• Vite build processing

Custom Video Game & AI Playground

PLAY

Phaser.js Vite AI Assets Modern JS

This self-built 2D game was developed using JavaScript, Phaser, and Vite. The core objectives were to learn game development principles and to explore the practical application of Large Language Models (LLMs) for generating game assets.

Tools like ChatGPT, Suno, and Gemini were utilized for AI-driven image, music, and video generation, complemented by custom sprite work in Photoshop.

Key Technical Learnings:

Phaser.js: Game loops, sprite animation, physics, state management.
LLM Integration: AI-driven asset generation (images, music, video).
Photoshop: Designed and animated custom 2D character sprites.
Deployment: Frontend build processes (Vite) and self-hosting (FTP).

Get In Touch

Let's connect.

Barcelona / Global Remote

I'm always open to discussing new data challenges, AI integrations, or technical leadership opportunities.

Email

kasperheyst@me.com

linkedin.com/in/kasperheyst

Send a Message

Kasper Thomsen

Journey

Senior Data Engineer

Scan Program Analyst

Data Analyst

Analytics Lead

Analytics Consultant

Technical Arsenal

Data Engineering

Analytics & BI

Cloud & ML

Education

MSc Business Intelligence

BSc Economics & Business Admin

BSc Business Administration

Key Projects

Smart Home Automation & Local LLM

Smart Home Automation & Local LLM

Global Scan Data Architecture (HP)

Global Scan Data Architecture (HP)

Custom Video Game & AI Playground

Custom Video Game & AI Playground

Get In Touch

Let's connect.