VaibhavMangroliya
Turning data into decisions.
Data Science • Machine Learning • AI • M.Sc. Mathematics
I write Python that goes to production. Right now that means a data-validation pipeline at LIST that gates Luxembourg's national environmental database. Before that, almost two years at India's National Stock Exchange: NAV calculation and XBRL parsing for 2,700+ listed companies.
Luxembourg 🇱🇺 · Open to relocation
Shipping production data pipelines at LIST · Esch-Belval
- 373
- tests at ~98% coverage
- 65+
- merge requests at LIST in 2 months
- 290K+
- YouTube views teaching what I learn
The thread?Turning data into decisions.
From a stock-exchange trading floor to a maths degree in Luxembourg to building intelligent systems: one throughline.
My journey
I started close to the markets, nearly two years at India’s National Stock Exchange, building production financial systems: a Python NAV-calculation tool across the Fair Value Hierarchy and the Oracle-SQL XBRL parser behind 2,700+ listed companies. Working that close to data is what pulled me toward the quantitative, model-driven side.
That pull brought me to the University of Luxembourg, where I’m finishing an M.Sc. in Mathematics with my focus firmly on machine learning, AI, and data. Alongside it I’ve built a real ML project portfolio (physics-informed neural networks, LSTM/XGBoost forecasters, and NLP with transformers), with the DeepLearning.AI foundations to back the theory.
Today at LIST I build DataQA, a production Python pipeline that ingests Luxembourg’s national environmental time-series, validates it against everything that can quietly go wrong, and loads the clean results back. Real data engineering: ETL, validation, scale, and the edge cases you only find when they break something, the foundation every good model is actually built on.
Looking for
Availability
Available from autumn 2026, once my current LIST internship concludes. Open to next steps in Luxembourg and across the EU.
Machine Learning & Modelling
Hands-on ML across a real project portfolio: physics-informed neural networks (PINNs) for option pricing at 99.8% accuracy, LSTM/XGBoost forecasting ensembles, and NLP with Hugging Face transformers. PyTorch, TensorFlow, scikit-learn, grounded in an applied-maths MSc and DeepLearning.AI's Supervised ML & Neural Networks courses.
Production Python & Data Engineering
Production Python at LIST: pandas, NumPy, pytest, ruff, REST APIs. Drove order-of-magnitude speedups by vectorising every check module, designed a Seasonal Check on per-(month, hour) statistical bands, and ship via GitLab merge requests. ETL, data validation, time-series at scale.
Quantitative Finance: a domain edge
Two years applying engineering to markets at NSE India: NAV computation, Fair Value Hierarchy (L1/2/3), VaR & Expected Shortfall, options pricing & the Greeks, GARCH. A domain I can speak fluently, and now apply machine learning and data to.
How I work
My standard is simple: code I won’t be embarrassed by in six months. At LIST that has meant order-of-magnitude speedups across the validation pipeline, eliminating CI hangs by hardening every external API call, and designing and shipping a new Seasonal Check module built on per-(month, hour) statistical bands. At NSE it meant a normalised 23-table SQL schema that cut data errors by 40%, instead of another fragile Excel workflow.
I also teach the things I learn. My YouTube channel has 290K+ views: Assembly, Engineering Physics, admissions guides. Explaining work to a non-expert audience is the same skill you need to write good client-facing reports.
Stack &expertise.
Machine learning and data engineering first, with the maths and quant-finance ground I have behind it.
Machine Learning & Deep Learning
Python & Data Engineering
Data & Databases
Quantitative Finance
Tooling & Languages
Currently Exploring
Where I’veshipped.
Two roles, one habit: write Python that other engineers (and regulators) can read.
Research Intern, ENVISION Unit (LEO Observatory)
InternshipCurrentLuxembourg Institute of Science and Technology (LIST)
- Maintainer of DataQA, a production Python pipeline (pandas, pytest, ruff) that validates Luxembourg’s national environmental time-series (air temperature, humidity, precipitation, global irradiance) against the KISTERS WISKI database via the KiWIS REST API. 65+ merge requests authored and ~200 merges reviewed in just my first 2 months.
- Built Global Irradiance validation from scratch: three new checks (physical-limit 0–1100 W/m² with night-time flagging via solar zenith angle, constant-value, and 20 km spatial-consistency), including the
solar_zenith.pyutility that derives sun position from station coordinates. - Designed and shipped a new Seasonal Check module for temperature and humidity, using per-(month, hour) statistical bands from historically validated data with a 90-day on-disk threshold cache.
- Vectorised every
iterrows()loop across all check modules for order-of-magnitude speedups, dropped NumPy as a direct dependency, and brought the codebase into pandas 3 / copy-on-write compliance. Hardened all KiWIS API calls with timeouts, eliminating CI hangs. - Grew the test suite to 373 functions at ~98% coverage with a skip-invariant harness and Ruff type-annotation enforcement. Built rank-correlating-stations, a companion pipeline producing per-parameter Pearson-correlation rankings (AT 0.94–0.99) across the 12-station network to feed DataQA’s spatial-consistency check, catching a glob-collision bug silently mixing 81 precipitation files into temperature analysis.
- Shipped a standalone River Discharge Hydrograph Generator CLI that renders print-resolution hydrographs from KiWIS data over per-calendar-day percentile baselines, built only from validated rows, with 100% test coverage and CI on Python 3.13.
Student Research Assistant, Department of Mathematics
Part-timeUniversity of Luxembourg
- Supported the Mathematics Department in preparing and typesetting research papers and academic documents in LaTeX (Overleaf), ensuring consistent formatting of equations, citations, and document structure.
- Assisted with the organisation of departmental events and academic activities, coordinating logistics and on-site support.
- Provided general academic and administrative support to faculty across the department.
Associate Systems Analyst
Full-timeNational Stock Exchange of India (NSE)
- Developed a Python-based NAV calculation tool automating Fair Value hierarchy classification (Level 1/2/3 assets), asset-liability aggregation, and Net Asset Value computation from Oracle database, directly applicable to investment-fund valuation and reporting.
- Built an XBRL parsing system transforming unstructured financial-statement data (Balance Sheet, P&L, Cash Flow) into a normalised SQL schema (23 tables). Reduced data errors by 40% and enabled automated validation across 2,700+ listed companies.
- Built Java / Spring Boot regulatory-compliance web applications enabling NSE’s compliance team to process SEBI filings, replacing manual Excel-based workflows with automated, audit-ready pipelines.
Academicbackground.
University of Luxembourg
M.Sc. in Mathematics
Mathematical Modelling & Computational Sciences
09/2024 – Present
Vidyalankar Institute of Technology
B.E. in Electronics & Telecommunication
Grade: 1.4, Top 4% in department
08/2018 – 05/2022
Certifications.
Click any card to expand and see the full topic coverage.
Selectedwork.
Things I built because I wanted to understand them properly. Machine learning, math, and data: usually all three.
Explore the Quant Lab
Live pricing tools, simulation notebooks, and quantitative finance experiments, running in your browser.
What people say.
“Vaibhav consistently stood out as a sharp and dependable professional. He showed a high level of ownership in his work, often handling critical modules with minimal guidance. Beyond his technical skills, Vaibhav is a collaborative team player with a professional attitude. I confidently recommend him for roles that require strong analytical thinking and problem-solving ability.”
Rahil Kamani
National Stock Exchange of India • 7.1 yrs exp.
Let’s talk.
Best for data science, machine learning, and data-engineering roles, with a Risk Management edge in finance. I read every message.