Open to Opportunities · UK

Ashish
Siwach

Data Analyst · Insights, Forecasting & Decision Science

MSc Business Analytics (Distinction) · University of Exeter · UK Graduate Visa

Built EV demand forecasting model achieving 15% WAPE, outperforming Prophet & LightGBM by 2×

Designed marketing attribution model (Adj. R²=0.92) identifying channels 2.5× more efficient, recommended 23% budget reallocation

Built RAG pipeline for climate misinformation detection, fine-tuning Llama-8B & Mistral-7B on 50 regulatory documents

View Projects LinkedIn GitHub

Ashish Siwach · Data Analyst

About

What I Do

I translate messy, complex data into decisions that move the needle, across experimentation, forecasting, attribution, and ML.

📈

Forecasting & Time-Series

SARIMAX, Prophet, LightGBM. Built EV demand models with 15% WAPE, outperformed two alternatives in head-to-head benchmarking.

🎯

Causal Inference & Attribution

Marketing mix modelling, A/B testing, Granger causality. Identified 2.5× channel efficiency gaps driving budget reallocation recommendations.

🤖

NLP & LLM Engineering

RAG pipelines, QLoRA fine-tuning on Gemma-2B, Mistral-7B, Llama-8B. Built climate misinformation detection system grounded on 50 regulatory documents.

🔬

ML & Experimentation

Scikit-learn, XGBoost, Random Forest. Delivered ROC-AUC 0.87 on healthcare classification with rigorous cross-validation and imbalance handling.

UK Graduate Visa holder based in Exeter. Available for analytics, data science, and ML roles across the UK. Currently seeking Data Analytics, Product Analytics, or Data Science roles across the UK.

University of Exeter

MSc Business Analytics — Distinction

Global Excellence Scholarship · 2024–2025

NMIMS Mumbai

Executive MBA, Business Analytics

CGPA 3.75 / 4.0

Experimentation Causal Inference Forecasting Optimisation Attribution Modelling Customer Behaviour A/B Testing NLP / RAG Python R · SQL Power BI

Projects

Work I'm Proud Of

9 end-to-end projects spanning forecasting, NLP, healthcare, and marketing, every one grounded in real data with measurable outcomes.

Experience

Work History

From AI research to operational leadership, building with data across industries.

Jun 2025 – Present
Exeter, UK

Part-time

Retail Assistant

Primark

Deliver customer service and floor support in a fast-paced retail environment.
Manage stock, replenishment, and merchandising while ensuring store standards.
Collaborate with team members to maintain smooth store operations during peak hours.

Customer ServiceTeamworkOperations

Jun – Dec 2025
Exeter, UK

Volunteer

AI & NLP Intern

University of Exeter — NCIT

Built end-to-end RAG pipeline for ClimaVAR, a climate misinformation analysis tool grounded on 50 official regulatory documents.
Engineered context optimisation workflows: query rewriting via CARDS API, chunking strategy tuning, document preprocessing.
Fine-tuned Gemma-2B, Mistral-7B, and Llama-8B using QLoRA; built evaluation pipelines with academic researchers.

PythonRAGQLoRA LLM Fine-tuningVector DBNLP

Sep 2014 –
Sep 2023
Sonepat, India

Operations Manager

R.K. Enterprises

Led a 12-member team; reduced material wastage by 47% through data-driven process analysis and operational redesign.
Monitored logistics and sales trends, achieving 87% repeat business rate through insight-driven customer retention.

OperationsTeam LeadershipProcess Optimisation

Skills