# Stuti Nayak

> ML Engineer and Data Engineer based in The Hague, Netherlands. Applied ML professional with 4+ years of industry experience plus academic research in machine learning, data pipelines and model evaluation. Builds transformer-based NLP and speech classifiers for noisy real-world data, LLM/RAG applications, and production data pipelines. Open to ML Engineer and Data Engineer roles in the Netherlands and EU.

- Website: https://stutinayak.com/
- Resume (PDF): https://stutinayak.com/StutiNayakResume.pdf
- Email: stuti257@gmail.com
- LinkedIn: https://www.linkedin.com/in/stuti-nayak/
- GitHub: https://github.com/stutinayak
- Location: The Hague, Netherlands
- Availability: open to roles in the Netherlands and EU

## Experience

### Data Engineer, Aurelius Enterprise (Amsterdam, Netherlands), Apr 2025 to present
- Supports a confidential client-facing AI automation initiative: examining source structures, preparing data and recreating operational KPIs to test technical feasibility.
- Designs evaluation criteria for AI outputs, defining success measures, data gaps, failure modes and stakeholder usability before model development.
- Designed and built production event and telemetry ingestion pipelines, transforming noisy upstream data into schema-validated datasets for analytics and downstream processing.
- Works with client and engineering stakeholders on data quality rules, dataset contracts, lineage and governance.
- Contributed feature development and bug fixes to an open-source metadata platform; documented integration decisions and technical handovers.

### Data Scientist, Knownwell (The Hague, Netherlands), Feb 2024 to Jul 2024
- Developed a BERT/BERTopic-based classification and topic-routing model, improving match accuracy from 20% to 30% on real-world text data.
- Built and validated a separate RoBERTa sentiment classifier using held-out evaluation and error analysis.
- Translated an ambiguous stakeholder question into a measurable ML problem with defined metrics and repeatable comparisons.
- Iterated on NLP preprocessing, inference prototypes and evaluation workflows; communicated trade-offs to non-technical stakeholders.

### Data Scientist, Rate.nl (Rotterdam, Netherlands), Aug 2021 to Sep 2023
- Built an audio classification pipeline using wav2vec 2.0 on noisy real-world call recordings, improving accuracy from 27% to 55% through preprocessing and experimentation.
- Designed and ran model experiments, tracked metrics with MLflow and analysed failure cases to choose a reliable production candidate.
- Packaged and deployed containerised model inference with Docker on Google Cloud Run.
- Built a Gradio application with model-output visualisations so stakeholders could inspect predictions and give feedback during rollout.

### Researcher, Clinical Data Science, Maastricht University (Netherlands), Sep 2019 to Aug 2020
- Configured federated learning infrastructure (Vantage) for privacy-preserving cross-hospital analytics.
- Designed a semantic model to standardise data across hospitals and reduce preprocessing overhead.

### Research Assistant, Institute of Data Science, Maastricht University (Netherlands), Jul 2017 to Feb 2019
- Built a deep learning NLP pipeline over scientific abstracts for biomedical metadata quality assessment.
- Developed CNN and BiLSTM multi-label classifiers; the CNN reached 83% accuracy on held-out evaluation.

## Skills

- Deep learning: Python, PyTorch, TensorFlow, Hugging Face Transformers, BERT, RoBERTa, wav2vec 2.0, CNN, BiLSTM
- ML evaluation: held-out validation, cross-validation, precision/recall, imbalanced learning, error analysis, statistical comparison, model selection, MLflow
- Production ML and software: scikit-learn, Docker, Google Cloud Run, Git, GitHub Actions, CI/CD, reproducible experiments, cloud inference, Gradio, FastAPI
- Data processing and pipelines: pandas, NumPy, SciPy, SQL, PySpark, Kafka, event and telemetry data, ETL/ELT, feature engineering, schema validation, data quality, BigQuery, PostgreSQL
- Cloud and applied AI: Google Cloud Platform, Azure, AWS, RAG, LangChain, FAISS, semantic search, confidence-based routing

## Selected projects

- Food QA Retrieval System: production-style Q&A API using Retrieval-Augmented Generation with document chunking and vector indexing (Python, FastAPI, LangChain, Transformers). https://www.datascienceportfol.io/stuti257/projects/1
- Credit Card Fraud Detection: LightGBM classifier on 284,807 highly imbalanced transactions; temporal splitting, threshold optimisation, ROC-AUC around 0.98, SHAP analysis.
- NLP Classification Pipeline: BERT/BERTopic multi-label classification and RoBERTa sentiment analysis pipeline (Python, PyTorch, Transformers, AWS). https://www.datascienceportfol.io/stuti257/projects/2
- Streaming Data Pipeline: IoT simulation pipeline with Kafka ingestion, a processing layer and ML analytics on Google Cloud (Python, Kafka, MLflow, GCP).
- Churn Prediction: end-to-end ML project predicting telecom customer churn, with SQL and behavioural feature engineering (Python, scikit-learn, pandas, SQL). https://www.datascienceportfol.io/stuti257/projects/0

## Education

- M.Sc. Data Science in Decision Making, Maastricht University, Netherlands
- B.Tech. Information and Communication Technology, Dhirubhai Ambani University (formerly DA-IICT), Gandhinagar, India

## Certifications

- Confluent Certified Developer for Apache Kafka
- Certified Data Management Professional (CDMP), Associate (DAMA International)
- Databricks Certified Data Engineer Associate (in progress)