← portfolio karan-homayounfar-cv
Karan Homayounfar
Data Scientist · Energy Markets & Time-Series ML · MSc Data Science
Forecasting GB electricity balancing costs with machine learning. Builds time-series pipelines from raw market data to evaluated models in production. First-class BSc graduate, enterprise scholarship holder.
karanhomayounfar2@gmail.com linkedin.com/in/karan-homayounfar github.com/knhnf karan-portfolio-al7.pages.dev
Bristol, United Kingdom
Right to work: UK Graduate visa eligible from 2027
Education
MSc Data Science, University of the West of England, Bristol
Jan 2026 - Jan 2027
Sem 1: Data Management Fundamentals 84 (Distinction) · Statistical Inference 70 (Distinction) · Dissertation: GB Balancing Mechanism Cost Forecasting
  • Student Representative, MSc Data Science cohort (~50 students) - Student Rep Gold Award 2026
  • UWE Enterprise Scholarship (£1,000) - local model phone reception system for SMEs
  • Member, UWE Student Society for Data and Computing · Bristol Data Week, University of Bristol 2026
BSc Computer Engineering 1st in graduating class GPA 18.87 / 20
2020 - 2024
Projects & Work
Forecasting total GB Balancing Mechanism cost per settlement period, a target no published GB study models directly. Pipeline ingests 131k+ half-hourly rows from the Elexon BMRS v2 API and NESO data, engineers 40+ features (imbalance volume, wind, demand, interconnectors), and compares a LEAR statistical baseline against XGBoost and LSTM under rolling-window out-of-sample evaluation with Diebold-Mariano testing. Due January 2027.
Reception System - Local Model Phone Handler In progress
2026 - Present
Offline voice assistant for SMEs: Faster Whisper STT, spaCy intent parsing, Piper TTS, FastAPI, Google Calendar API. Fine-tuned Phi-3 and Llama 3.2 with QLoRA/PEFT adapters to 98 to 99.8% action accuracy, evaluated on a 480-sample, 4-condition suite (WER, CER, latency). UWE Enterprise Scholarship.
Bristol Air Quality - Relational and Document DB Analysis
2026
Dual-paradigm analysis. MySQL with PHPMyAdmin/Workbench, MongoDB aggregation pipelines. MSc Distinction (84).
End-to-end BI pipeline. ETL with Python and Pandas, MySQL storage, Matplotlib dashboards.
Reads engineering quote emails, extracts structured data, drafts replies. Ollama local models + OpenAI GPT. 96% field accuracy.
Production site for a dental practice. Cloudflare Pages, GA4, GTM, Clarity, Web3Forms, before/after sliders, dark mode. Live and receiving patient enquiries.
Skills
Languages
Python, R, SQL, JavaScript, HTML/CSS
Data & ML
Time-series forecasting, XGBoost, Scikit-learn, Pandas, NumPy, LLM fine-tuning (QLoRA/PEFT), spaCy NER, Faster Whisper, Piper TTS, model evaluation (WER/CER, latency), R (tidyverse, ggplot2)
Databases
MySQL, PostgreSQL, MongoDB
Deploy & Infra
Cloudflare Pages, Render, Vercel, Supabase, FastAPI, Streamlit, Docker, GA4, GTM
Developing
SHAP · LangChain · RAG · C++ · C# · GitHub Actions · n8n · Clarity · Power BI
Achievements & Activities
1st in BSc graduating class, GPA 18.87/20
UWE Enterprise Scholarship holder
MSc Student Rep Gold Award 2026
Bristol Data Week, University of Bristol 2026
UWE Hackathon 2025 & 2026
Five production sites built, deployed, and live