Nottingham, UK
Data Scientist, MSc Data Science, University of Nottingham
Predictive modelling in Python and R, statistical machine learning, reproducible analysis
I am a data scientist with an MSc in Data Science from the University of Nottingham and a first degree in mathematics from the University of Delhi. I build predictive models, test them with paired experiments and nested validation, and explain the results to people who are not statisticians.
For my dissertation I implemented Mondrian forests from scratch in Python. In summer 2026 I delivered a live data product for a San Francisco start-up in two weeks.
06/2026 – 07/2026
Hermai.ai, client project
A live site that shows how much sleep each of the 48 fanbases loses to the kick-off times of the 2026 World Cup. I cleaned the 104-match fixture data, mapped every fanbase to its home time zone and set up a six-hourly rebuild that halts publication if any fixture fails its checks.
Spring 2026
Statistical Machine Learning, group project (Distinction)
A group project that predicted the impurity type and percentage of 100 chemical samples. Our R pipeline placed first of three groups on the prediction leaderboard. Replacing a random forest with one linear model per impurity type reduced the error score by 33%.
2025
Databases coursework (Distinction)
A hospital records system on MariaDB. I extended a 7-table schema to 9 tables, joined each patient’s admissions, tests and examinations into one searchable view, and added an audit log of 8 kinds of user action.
2025
Data science project
Sentiment scoring and a 10-topic model on 2,000 Skytrax reviews put cancellations and refunds at 20.7% of all reviews. Tree-based models on 50,000 bookings ranked purchase lead time, route and flight hour as the three strongest signals.
2025 – 2026
Time Series and Forecasting
Six-month forecasts with 80% and 95% prediction intervals from two seasonal ARIMA models fitted to ten years of monthly data, with a one-page summary written for non-technical readers.
My research interests are statistical machine learning, computational statistics and the behaviour of methods when the training data are contaminated.
2026
MSc dissertation, marked at 80%
An empirical study of Mondrian forests against CART and random forests under changes in sample size, label noise and feature-space outliers. I implemented the method from the algorithms of Lakshminarayanan, Roy and Teh (2014) and ran every comparison on paired seeds. Choosing the lifetime by validation reduced excess error by a factor of 7 to 23.
Spring 2026
Research paper, Data Science with Machine Learning
Four classifiers compared on 29,628 police collision records from the East Midlands with nested cross-validation (10 outer folds, 3 inner). Random forest had the highest macro-F1 (0.555 ± 0.014) and came first in all ten outer folds. The lecturer chose the analysis as a teaching example for the 2026–27 cohort.
Autumn 2025
Group report (team of four), Statistical Machine Learning
I wrote the mathematical sections, from the graph Laplacian to the relaxations of RatioCut and the normalised cut as eigenvalue problems, with every step worked on an 11-node example graph. I also wrote the discussion and most of the conclusion.
almost surely (opens in a new tab)
Definition. An event happens almost surely if it has probability one.
My notebook on statistics, machine learning and mathematics: methods built from the ground up, problems worked in full, and notes on AI and the news, with the working shown.
In the notebook
06/2026 – 07/2026
Hermai.ai, University of Nottingham micro-placement
Delivered the World Cup Sleep Index for the co-founder in two weeks, as the analyst on that project in a seven-person team.
05/2024 – 07/2024
The Sparks Foundation
Analysed monthly sales trends and compared performance across stores and product categories using SQL JOINs, CTEs and aggregate functions. Cleaned and reconciled sales and product records in Excel using Power Query, XLOOKUP and PivotTables. Developed interactive Power BI dashboards with DAX measures and slicers to track sales performance and consolidate monthly reports.
2019 – 2025
Self-employed, India
Taught students aged 11 to 18, one to one and in small groups, from algebra to calculus.
09/2025 – 09/2026
University of Nottingham
Statistical Machine Learning 80%, Data Science with Machine Learning 78%, dissertation 80%. South Asia Postgraduate Excellence Award.
06/2022 – 05/2025
Sriram’s IAS, Delhi, India
Full-time preparation across three annual cycles for India’s Civil Services Examination.
06/2019 – 07/2022
University of Delhi, India
CGPA 7.53/10. INSPIRE scholarship from the Government of India for a top 1% result in the Class XII board examinations.