WELCOME
Devika Rajasekar
Data & Analytics Engineer
Netherlands · She/her

“Why?” is my favorite data point.

I work on data infrastructure and machine learning models. Also, I will try any gelato flavor once, and I bring the same energy to a new tool or dataset. I learn it properly, get obsessed, build something with it. That’s basically the story behind every project below.

Download CV See my work
Devika Rajasekar professional data nerd, part-time ice cream critic
GitHub LinkedIn Email 
⠀⠀⠀⠀⢠⡶⠚⢷⣤⡀⠀⠀⠀⠀⠀⣲⡶⠛⠻⣆⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⢠⡿⠁⠀⠀⠙⣷⣄⠀⢀⣴⡟⠁⠀⠀⢷⢹⡆⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⠀⣾⠃⠀⠠⠶⠚⠛⠛⠛⠛⠋⠀⠀⣀⡀⢸⠈⣿⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⢸⣏⡔⠋⠀⠀⠀⠀⠀⠀⠀⠀⠀⠚⠉⠉⣿⠀⢹⠀⠀⠀⠀⠀⠀⠀⠀⠀
⠀⠀⢾⠏⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠸⠀⢸⡇⠀⠀⠀⠀⠀⠀⠀⠀
⠀⢠⣿⢠⣶⡆⠀⠀⠀⠀⣀⣀⠀⠀⠀⠀⠀⠀⠀⠀⢸⡇⠀⠀⠀⠀⠀⠀⠀⠀
⢒⡾⠁⠘⠟⠁⠀⠀⠀⠀⣿⣿⡆⠀⠀⠀⠀⠀⠀⠀⢸⡇⠀⠀⠀⠀⠀⠀⠀⠀
⠉⣧⠀⠀⠀⠀⠃⠀⠀⠀⠈⠉⠠⣍⠀⠀⠀⠀⠀⠀⣸⡇⢀⣤⠶⠛⠛⠻⢦⣄
⠀⠸⣧⡀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⣰⡟⣴⠟⠁⠀⠀⠀⠀⠀⢻
⠀⠀⠀⠛⣷⡦⠀⠀⠀⠀⠀⠀⠀⠀⣀⣀⣤⡴⠞⠋⢠⡟⠀⠀⠀⠀⠀⠀⢀⡾
⠀⠀⠀⢰⡿⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠉⠳⣤⡀⢸⠃⠀⠀⠀⠀⢠⡶⠟⠁
⠀⠀⠀⣸⠇⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠘⢷⣹⡄⠀⠀⠀⠀⣼⠀⠀⠀
⠀⠀⠀⣿⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⢿⣇⠀⠀⠀⠀⢹⡄⠀⠀
⠀⠀⠀⢸⡀⢀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⣿⡄⠀⠀⠀⠈⣧⠀⠀
⠀⠀⠀⢸⡇⠘⡇⠀⠀⠀⠀⠀⠀⠀⣀⠀⠀⠀⠀⠀⠀⢸⣿⠀⠀⠀⠀⢹⡇⠀
⠀⠀⠀⢸⡇⠀⠙⠀⠀⠀⠀⠀⢠⠞⠁⠀⠀⠀⠀⠀⠀⠀⣿⠇⠀⠀⠀⢸⡇⠀
⠀⠀⠀⢸⡇⠀⢸⡆⠀⠀⠀⠀⣟⠀⠀⠀⠀⠀⠀⠀⠀⠀⠛⠀⠀⠀⠀⣸⠇⠀
⠀⠀⠀⢸⣿⠀⠀⡇⠀⠀⠀⠀⣿⡀⠀⠀⠀⠀⠀⠀⠀⢀⡇⠀⠀⢀⣴⡟⠁⠀
⠀⠀⠀⠘⠿⠶⢶⢧⣦⣦⡴⢾⣥⣽⣤⣤⣤⣤⣤⣤⡴⣯⡤⠴⠶⠛⠋⠀⠀⠀
ps. drink some water
psst, the stickers moveTRY DRAGGING ME
Meet the engineer

Hello,
I’m Devika!

I recently graduated with my masters in Computer Science (Data Science) from Leiden and finished an HR Digital Data Science internship at Prysmian in Delft. I like being in all the parts of a data problem: shaping raw tables, building models, judging when their output can be trusted, and getting a result into a form people can act on.

The human side of things has always pulled at me, sociology, anthropology, biology, so taking my CS degree alongside marketing and lean-startup courses felt natural. Who reads the output, and what do they change because of it? It’s the piece I look for first in any new project.

off the clock, I’m all about

cinema ticket, movies & shows coffee stamp spicy food stamp
Devika eating gelato
the best gelato I’ve had in my life
Where I’ve worked

Experience

2025AUG 2025 – AUG 2026

HR Digital Data Science Intern

Prysmian · Netherlands
  • Replaced monthly manual KPI consolidation across 53 plant-level Excel sheets with a single live Qlik Sense dashboard fed from a central data lake, covering 15 countries and 4 regions (CEE, NE, SE, UK), trusted by senior stakeholders.
  • Translated cross-functional reporting requirements into reusable data models, KPI definitions and dashboard logic covering variance analysis and conditional indicators for senior leadership.
  • Completed my MSc thesis on AutoML explainability: whether SHAP feature-importance stays consistent across the Rashomon set of near-optimal models, with AutoGluon & H2O experiments deployed on an HPC cluster.
Qlik SensePower AutomateAutoMLSHAPHPC
2023MAY – JUL

Deep Learning & Image Processing Research Intern

CCPS, VIT Chennai · India
  • Built a custom U-Net variant (AdaptUNet) to detect colorectal polyps from 2,000+ medical images: 0.9104 Dice, 0.9880 balanced accuracy.
  • Ran cross-dataset generalisation and data-validation checks. Published as first-author in Elsevier Heliyon (2024) ↗.
PyTorchU-NetMedical Imaging
2023JUL – NOV

Data Analysis Intern

School of CSE, VIT Chennai · India
  • Designed a deep neural network in R for pulsar-star classification: 98% precision, 0.8749 Kappa on the HTRU2 dataset.
  • Led reproducibility & compliance checks to IEEE academic QA standards, underpinning a successful conference publication ↗.
RDeep LearningIEEE
2022AUG – DEC

Research Assistant

School of CSE, VIT Chennai · India
ResearchRoboticsLiterature Review
whimsical illustrated cat covered in colourful bows
A pinboard of projects

Some of my work

Data Engineering

paytrail: Payments Medallion Lakehouse

  • Built a Databricks medallion lakehouse taking 6.3M synthetic payments from Azure storage through bronze, silver and gold layers.
  • Designed idempotent incremental loads for late and out-of-order data, with Unity Catalog governance and 51 dbt tests gated in CI.
  • Reconciled settlement outputs to source at penny level and reduced the heaviest rollup runtime by 39% after optimisation.
DatabricksDelta LakeUnity CatalogdbtAzureGitHub Actions
GitHub ↗Read the writeup ↗
Data Engineering

Cross-System Customer Data Reconciliation

  • Reconciled customer records across Salesforce, NetSuite and Stripe-style data, starting from 12,566 supplied case rows.
  • Built deterministic exact-key cleaning for duplicate records, missing identifiers, mixed date formats and European/US currency formats.
  • Kept unresolved links visible instead of inferring them: 49 NetSuite rows missing a Stripe ID and 675 unmatched Stripe transactions surfaced as exceptions.
PythonpandasDeterministic Matching
GitHub ↗
Data Engineering

Workplace Safety Analytics Pipeline

  • Built a batch pipeline ingesting 688K+ OSHA injury records into a PostgreSQL star schema, orchestrated through a seven-task Airflow DAG.
  • Added local LLM enrichment for free-text incident narratives using Ollama, keeping the pipeline reproducible and inference cost-free.
  • Enforced 81 automated data and contract tests through dbt and GitHub Actions, with failed checks blocking the build.
Apache AirflowdbtPostgreSQLDockerGitHub ActionsOllamaPython
GitHub ↗
ML & NLP

The Edit: H&M Recommender

  • Built a two-stage recommender over 31M H&M transactions, using BigQuery SQL retrieval followed by a CatBoost ranking model.
  • Improved MAP@12 from 0.0053 to 0.0292, with a 95% bootstrap confidence interval confirming the lift was stable.
  • Added a temporal-leakage gate, diversity checks, 80+ automated tests, and a typed FastAPI prediction endpoint.
BigQueryCatBoostFastAPIPythonSQL
GitHub ↗
Forecasting & Research

SHAP Stability Across the Rashomon Set

  • Tested whether SHAP feature rankings remain stable when a forecasting model is replaced by another near-optimal model.
  • Ran AutoGluon and H2O experiments across six forecasting benchmarks, multiple temporal splits and Rashomon thresholds on Leiden's ALICE HPC.
  • Found H2O built larger, more varied Rashomon sets than AutoGluon (up to 15 models across 5 families, against up to 5 in 1), yet both frameworks' stability agreed within 0.04 Spearman rho on every dataset (0.77 to 0.98 overall).
AutoGluonH2OSHAPPython
GitHub ↗Read the writeup ↗
Forecasting & Research

Rossmann Store Sales Forecasting

  • Built a LightGBM forecasting pipeline for 1,115 Rossmann stores across roughly 1.02M historical rows.
  • Used expanding-window validation with train-only store aggregates to keep the evaluation leakage-safe.
  • Reached 11.6% mean RMSPE against a 24.7% store-weekday median baseline; these are local validation results, not a Kaggle leaderboard score.
LightGBMOptunaPythonstatsforecast
GitHub ↗
ML & NLP

Two-Tower Movie Recommender

  • Built a two-stage recommender over 25M MovieLens ratings, using PyTorch Two-Tower retrieval with FAISS followed by LightGBM ranking.
  • Diagnosed embedding collapse (cosine similarity to the centroid measuring 1.000) and fixed it by swapping BatchNorm for LayerNorm, taking Recall@10 from 0.002 to a final 0.147.
  • Evaluated the engagement/diversity trade-off offline through a Pareto frontier and SNIPS counterfactual evaluation.
PyTorchFAISSLightGBMPython
GitHub ↗Read the writeup ↗
Data Engineering

Lead Conversion Data Product

  • Designed a relational schema for lead-to-member conversion and containerised the Postgres database with Docker.
  • Built a dashboard prototype so business managers can explore 4 revenue KPIs directly.
PostgreSQLDockerPythonSQL
GitHub ↗
ML & NLP

NER in the CSIRO Adverse Drug Event Corpus

  • Fine-tuned BioBERT to extract adverse drug reactions from noisy biomedical text on the CSIRO CADEC corpus.
  • Used Focal Loss to counter severe class imbalance between common and rare entity types.
  • Lifted rare-entity F1 by over 20 points while reaching 88.92% overall accuracy (F1 88.04%).
BioBERTPyTorchHugging FaceFocal Loss
GitHub ↗
Data Engineering

Fashion Analyzer

  • Built PySpark pipelines over 2 years of fashion sales data spanning 9 categories, 10 styles and 5 regions.
  • Aggregated by category, style and region to surface assortment signals rather than raw transaction counts.
  • Found mid-range pricing ($50 to $150) drives the highest volume, with style preferences splitting sharply by market.
PySparkPythonData Viz
GitHub ↗
Forecasting & Research

Energy Time-Series Forecasting for Hydropower

  • Forecast hydro generation across Eastern India from 16 months of National Power Portal data.
  • Selected between ARIMA, SARIMA and Prophet using AIC and BIC rather than picking a model by eye.
  • Best model reached an RMSE of 153.55 and MAE of 78.03 on held-out months.
SARIMAARIMAProphetPython
GitHub ↗
Forecasting & Research

Reinforcement Learning Benchmarks

  • Implemented REINFORCE, Actor-Critic and A2C policy-gradient algorithms on CartPole-v1 in TensorFlow.
  • Extended the environment with transaction-cost penalties to simulate real-world P&L drift.
  • Ran 5 independent 1M-step experiments per method and compared final reward stability, not just peak reward.
TensorFlowPythonOpenAI Gym
Forecasting & Research

Graph Anonymization: Topology Preservation vs Privacy

  • Benchmarked 4 anonymization methods across 5 real-world graphs for structural preservation and re-identification risk.
  • Measured community-structure degradation separately from degree-distribution degradation, rather than a single blended score.
  • Found modularity shift stayed under 3% and re-identification risk under 1%, but aggressive anonymization degrades community structure faster than degree distribution.
PythonNetworkX
GitHub ↗
Data Engineering

where-the-money-goes-next

  • Built a fraud-detection pipeline scoring accounts most likely receiving money from authorised-payment scams, across 6.9M synthetic transactions.
  • Benchmarked a hand-written rule against logistic regression and XGBoost rather than assuming the ML model would win.
  • Set the alert threshold to a fixed daily analyst-review budget, with 173 tests gating the pipeline including a temporal-leakage check.
PythonpandasNumPyscikit-learnXGBoostSciPy
GitHub ↗
Forecasting & Research

Bayesian Media Mix Model for Advertising Spend

  • Built a Bayesian media mix model estimating each advertising channel's contribution to sales from two years of weekly data.
  • Ran a controlled comparison isolating spend burstiness as the reason the saturation knee goes unlearned on most channels.
  • Validated the PyMC inference with simulation-based calibration (15 of 15 parameters passed) and 139 automated tests.
PythonPyMCPyTensorArviZNumPypandas
GitHub ↗
View all my projects by category →
My toolkit

Tools & skills

Languages

PythonRSQL

ML & Deep Learning

PyTorchTensorFlowScikit-LearnTransformersComputer VisionPySparkReinforcement Learning

Data Engineering

DatabricksDelta LakeUnity CatalogdbtAirflowDatabricks Asset BundlesDockerGitUiPath

Analysis & Viz

PandasNumPyMatplotlibSeabornPower BITableauQlik SenseExcel

Cloud

Microsoft Azure (ADLS Gen2)AWS (S3 · EC2 · Lambda)Google Cloud

Ways of working

Agile / ScrumData ValidationRequirement GatheringUser Story CreationStakeholder CommsAutomated Testing
On paper

Background

MSc Computer Science: Data Science
Leiden University, Netherlands
SEP 2024 – AUG 2026
Coursework: Automated Machine Learning · Recommender Systems · Reinforcement Learning · Social Network Analysis · Text Mining · Data Mining. Thesis: Stability of SHAP explanations across near-optimal AutoML models for time-series forecasting.
Bachelors of Technology Computer Science Engineering: AI & Robotics
Vellore Institute of Technology (VIT), India
SEP 2020 – JUN 2024 · CGPA 8.5 / 10
Machine Learning · Linear Algebra · Deep Learning · Data Analysis, sociology, marketing & lean-startup.

Publications

Efficient Colorectal Polyp Segmentation Using Wavelet Transformation and AdaptUNet: A Hybrid U-Net ↗
Elsevier Heliyon · 2024 · first author

Built AdaptUNet, a custom U-Net combining attention mechanisms with wavelet features for colorectal polyp segmentation. Trained on Hyper-Kvasir and evaluated on three external datasets, reaching 0.9104 Dice and 0.9880 balanced accuracy on CVC-300.

A Review Paper on Mobile Robots Applications in Search and Rescue Operations ↗
Advances in Science and Technology · 2023 · co-author

Defined the inclusion criteria, built the comparison taxonomy, produced the figures, and wrote the review across 20+ studies on mobile robots for search-and-rescue operations. Presented at ICoFT MADE 2022.

A Machine Learning Approach for the Identification and Estimation of Pulsars upon Statistical Analysis ↗
IEEE NMITCON · 2023 · co-author

Built and tuned a deep neural network in R for pulsar classification on HTRU2, then verified the reproducibility of the evaluation. The final model reached 98% precision and 0.8749 Kappa.

Languages

EnglishNATIVE
DutchA2 niveau
Hindi & TamilFLUENT

Leadership & engagement

General Secretary · CodeChef VIT Chennai
JAN – JUL 2023

Drove recruitment to 500+ applicants. Ran coding events, workshops and seminars to grow a vibrant coding culture.

Active Member · IEEE Women in Engineering, VIT
MAY – DEC 2021

Advocated for women in STEM and organised professional-development events across diverse stakeholder groups.

Licenses & certifications

View all on LinkedIn ↗

AWS: Introduction to Machine Learning on AWS · Google Cloud: Big Data & Machine Learning Fundamentals · IBM: Artificial Intelligence Analyst

View my CV ↗