Lorenz attractor: deterministic equations that model chaotic systems. The mathematical origin of the butterfly effect.
Hello, I'm
Natan Szczepaniak
Senior Data & AI Consultant
5+ years experience bridging data science & human-centred design building end-to-end AI/ML solutions across healthcare, financial services, retail, automotive and more. MPhys physicist turned consultant.
Recent Posts
ML Interatomic Potentials
Machine Learned Interatomic Potentials for MD simulations
Physics
University
ML
X-Ray Crystallography
Investigation into the crystal structure of Lithium Fluoride, Potassium Bromide and quartz.
Physics
University
Neuromorphic Computing
Neuromorphic Computing for Artificial Intelligence
Physics
University
Skills
Programming & Web
Linux
Python
SQL
JS/TS
HTML
CSS
Bash
REST APIs
FastAPI
Flask
React
Node.js
Next.js
Streamlit
Git
Nginx
AI & Machine Learning
ML
Deep Learning
NLP
LLMs
MLOps
Agentic AI
RAG
ReAct Agents
PyTorch
TensorFlow
scikit-learn
Hugging Face
LangChain/Graph
MLflow
SageMaker
Pandas
Vector Databases
Cloud & Data Engineering
AWS
GCP
Linux
Docker
Terraform
CloudFormation
CI/CD
Jenkins
Airflow
Spark
Snowflake
PostgreSQL
MongoDB
Redis
PowerBI
Tableau
Salesforce
ServiceNow
Consulting
Stakeholder Management
Business Development & Sales
Workshop Facilitation
Technical Discovery
Project Scoping & Proposal Writing
Roadmap & Strategy
Cross-functional Collaboration
Agile
Jira
Confluence
Career
2023
2023 – Present
Senior Data & AI Consultant
Method (a GlobalLogic Company), London
- Architected a phased digital transformation roadmap for Hitachi Global Life Solutions, designing a CX data capture solution for sales staff across Japan, mapping data flows into reporting and forecasting models for leadership. Presented to C-suite executives in Tokyo; later showcased at the Hitachi Digital Summit in San Francisco. Contributed to $9.2M in additional tracked annual revenue.
- Developed a custom HIPAA-compliant LLM health insurance plan tradeoff model for a leading Medicare broker in the US, using information theory based feature selection surfacing most relevant plan attributes before summarising using a self hosted open source LLM using AWS SageMaker containers translating raw plan data into personalised natural language summaries. Additionally, designed the event analytics data layer tracking user behaviour and interactions across the digital experience to track adoption and perform customer segmentation.
- Designed and developed 8 ML model POCs for a major EdTech platform optimising the full student lifecycle including lead scoring, collaborative filtering cross-selling, CLTV, pre-enrollment qualification and mid-course risk detection alongside monitoring dashboards.
- Independently scoped, designed and built a Power BI cybersecurity vulnerability monitoring dashboard combining application and network scanner data across ~300 products for FIS Digital. Became the 43rd most viewed dashboard in the organisation in 3 months, with the quality of delivery leading to an 8-month extension worth $267K.
- Co-led a feasibility assessment for Subaru America, identifying and evaluating high-value advanced analytics use cases using vehicle telematics data sources to improve brand loyalty, connected customer experience and retention.
- Co-authored From Experiments to Enterprise: Industrialising AI through MLOps article for method.com and featured on the accompanying Build What's Next podcast discussing MLOps best practices. Delivered multiple internal thought leadership presentations on topics including Demystifying GenAI, Colour Theory, Streamlit, and project case studies.
- Built two internal tools for the Data & AI team: a synthetic data generation platform used to accelerate prototyping and a multithreaded web scraping tool used to analyse client website structure and content ahead of engagements.
- Built an interactive iPad web app for the FIS Emerald Convention 2024 showcasing customer segments, used by hundreds of banking executives at the financial convention in Florida.
- Created reusable frameworks and accelerators adopted across the team including an MLOps Agnostic Framework, Golden EDA template and Streamlit accelerator. Managed Data & AI team's AWS and GCP accounts and built internal Slack bots to automate team processes.
- Helped scale the Data & AI team from 4 to 16 members over 3 years, reviewed 230+ applicant profiles, conducted 30+ interviews, created the technical hiring assessment, and mentored junior data designers and interns.
- Supported sales of over $1M in completed projects across various industries through scoping, writing SOWs, workshops, discovery sessions, and POCs. Awarded Gold and Silver Commercial Recognition two consecutive years.
2021
2021 – 2023
Data Scientist / MLOps Engineer
GlobalLogic, London
- Embedded within NatWest Group's MLOps platform team alongside AWS Professional Services, serving as an SME to onboard data science teams onto a new AWS SageMaker-based platform. Built reusable SageMaker Pipeline templates for model deployment and advised teams on best practices and solution design, helping reduce idea-to-value time from 40 to 16 weeks across the organisation.
- Supported the productionisation of 8+ ML models including fraud detection, CLTV, NLP-based customer service and NPS analysis.
- Led a team of 4 junior developers to build an MVP for a full stack (React & Flask) application which calculated the scope 3 carbon emissions of AWS resources using Cost Explorer API and formulas for calculation. The application was used by the sales team to drive more sales in the ESG sector before the release of AWS Customer Carbon Footprint Tool.
- Developed an automated phone number verification system using Python and Twilio for a major telecoms provider saving ~£100k in marketing costs.
2018
2018
Data Science Intern
Applegate Marketplace, Devon
- Designed and implemented a ML lead scoring model for a B2B procurement marketplace with 140,000+ suppliers, scoring leads based on source channel, interaction frequency, and engagement signals to help sales teams prioritize outreach. Built as part of a data science collaboration between Applegate and the University of Exeter.
2017
2017 – 2021
MPhys Master of Physics (2:1)
University of Exeter, Devon
Dissertation: Machine Learned Potentials for Molecular Dynamics Simulations (79%). Coursework spanning quantum mechanics, astrophysics, condensed matter physics, and scientific programming in C.
Awards & Certifications
Awards
- Silver Method Commercial Award
- Gold Method Commercial Award
- Best of Method Award
Certifications
- AWS Certified Solutions Architect Associate
- AWS Cloud Practitioner
- European Economic and Social Committee — Pathways to Data Analytics