Hi, I'm Daria —
where biotech meets machine learning.
Turning complex biomedical data into something that helps people.
I build data-driven AI systems with real-world impact — from LLM-powered applications to biomedical signal and image analysis. I'm happiest in places that value curiosity, bold ideas, and work that quietly makes the world a little better.
A scientist's mind, a builder's hands.
I'm a biomedical engineer with a strong foundation in mathematics, machine learning, and AI. My work lives at the meeting point of biology and code — building systems that learn from messy, real-world data and turn it into something useful.
My experience spans biomedical signal and image processing, LLM-based systems, and quantitative modeling — so I'm just as comfortable analysing an ECG or an MRI as I am designing an AI pipeline or pricing a derivative. That range lets me move easily between deeply technical and applied problems.
What truly drives me is work that meaningfully improves the world — especially in health and deep tech — and teams that make room for curiosity, unconventional thinking, and a long-term vision.
Two worlds, one toolkit.
My heart is in biotech and machine learning — and everything I build sits somewhere between the two.
Machine Learning & AI
Supervised & unsupervised learning, neural networks (ANN, CNN), feature engineering, dimensionality reduction, and robust statistics — from first idea to a tuned, validated model.
Biomedical Signal & Imaging
ECG, EEG, EMG and sleep signals; MRI, fMRI, CT and ultrasound. Biomarker detection, disease-progression modeling, and turning physiology into reliable, actionable insight.
LLM-Powered Systems
Designing and scaling production AI: LLM pipelines, structured output extraction, conversational voice assistants, and workflow orchestration with LangChain, n8n and Vapi.
Quantitative Modeling
Derivatives pricing, market-risk analysis, and stochastic optimisation in Python and R — structured, analytical problem-solving under real uncertainty.
Where I've been making things.
- Price exotic and vanilla derivatives in Python and R, and manage market risk through robust quantitative analysis.
- Designed a stochastic optimisation algorithm for commodity trading that improved strategy performance under uncertainty.
- Present quantitative results to senior stakeholders, translating complex models into clear decisions.
- Owned performance optimisation and scaling of production LLM-powered voice assistants.
- Led training and deployment pipelines with n8n and Vapi for cost-efficient, human-free operations at scale.
- Optimised an AI pipeline for importing blood-test data (OpenAI & Anthropic APIs, Google Cloud Vision OCR).
- Improved extraction logic and meaningfully reduced manual processing.
- Built the experimental design and ML model for detecting volatile organic compounds (biomarkers).
- Reached 90%+ accuracy with supervised & unsupervised methods, feature selection and hyper-parameter tuning.
Things I'm proud of.
A few projects where biology, signals and machine learning came together.
Sensing Gases & Volatile Biomarkers
An experimental design and ML model to identify volatile organic compounds with chemo-resistive sensors, across varying conditions and concentrations. Included in a funded grant for other researchers.
Early Parkinson's from Speech
A machine-learning model detecting early Parkinson's disease from voice — built on careful feature selection, hypothesis testing and classification. Won 2nd place at a CTU data-analysis competition.
Medical Image Processing
Analysed MRI with SPM12, processed and filtered ultrasound, worked with CT scans, and developed segmentation algorithms for microscopic images to analyse tumor cells.
Safety Device for Firefighters
A prototype wearable using an ECG sensor with R-peak detection for heart-rate monitoring — alarming if heart activity dropped — plus a CO sensor to catch dangerous carbon-monoxide levels.
CNN for Image Classification
Designed and trained a convolutional neural network in PyTorch, tuning learning rate and batch size to optimise classification performance.
Employee Attrition Modeling
Predictive modeling on the IBM attrition dataset — EDA, stepwise feature selection, dimensionality reduction and classification with logistic regression and k-NN.
Skills & tools I love working with.
Machine Learning & Data Science
AI & LLM
Biomedical Signal & Image
Programming
Mathematics for ML/AI
Tools & Platforms
I'd love to hear from you.
Whether you're building something in biotech, machine learning, or somewhere wonderfully in between — let's talk about how I can help.