A-IT Support Resilient Ticket Assistant
An LLM-powered agent that orchestrates ticket collection, classification, and grounded responses so support teams can close loops faster.
The Data Science Journey
Here is my journey in applied machine learning, analytics, and physics-driven research built for real-world impact.
From agentic support assistants to cosmological model testing, I connect six years of quantitative research with pragmatic product thinking. Explore the highlights below to see how I design, ship, and interpret data-centric solutions. Welcome! Explore the highlights below.
Django & ArXiv Article Discovery
A full-stack filter that ranks ArXiv submissions with NLP, surfacing the papers that matter to a target research agenda.
Periodic K-Means on Household Power Data
Clustering household energy signatures with a periodic kernel to reveal consumption archetypes and demand peaks.
Biodiversity EDA & Hypothesis Testing
Exploratory analytics for U.S. National Parks biodiversity, pairing statistical testing with storytelling visuals.
Forest Cover Type Classification
Sequential Keras models that push predictive accuracy on forest cover labels while balancing complexity and generalisation.
Stack: Keras, TensorFlow, structured feature engineering, disciplined experiment tracking.
The notebook walks through the model family and highlights how additional depth and regularisation impact performance.
Cosmology Model Testing (NEDE)
Bayesian inference for novel dark energy models, blending theoretical physics priors with observational data constraints.
Fundamentals of NLP – Disaster Tweets
End-to-end tweet classification with conventional ML pipelines, showing that solid features can rival deep models on lean datasets.
Workflow includes data cleaning, tokenisation, TF-IDF, Naive Bayes, Logistic Regression, and disciplined hyper-parameter tuning.
