Available for work

Karan Homayounfar

MSc Data Science · Bristol. I build pipelines, models, and web systems that run in production - not just notebooks.

karan@bristol ~ profile.json
// Karan Homayounfar
{
  "role": "Data Scientist & Developer",
  "degree": "MSc Data Science, UWE Bristol",
  "bsc_rank": "1st in class", // GPA 18.87/20
  "grades": [84, 70], // sem 1
  "focus": [
    "ML Engineering",
    "Data Pipelines",
    "Automation Tooling"
  ],
  "open_to": "internships & graduate roles"
}
work

Projects

📞
Reception System
Fine-tuned local LLM phone handler for SMEs. Faster-Whisper STT, QLoRA fine-tuning, 99.8% action accuracy. UWE Enterprise Scholarship - prototype, no live client yet.
PythonQLoRAFaster WhisperFastAPI
🤖
SmartQuote Assistant
Reads engineering quote emails, extracts structured fields, drafts replies. Ollama local + GPT fallback. 96% field accuracy.
PythonOllamaOpenAI
📊
Retail Sales KPI Analysis
End-to-end pipeline over £10.6M dataset. ETL with Python and Pandas, MySQL storage, Matplotlib dashboards. BI-style workflow.
PythonMySQLPandas
GB Balancing Mechanism Forecasting
MSc dissertation. Forecasting total GB electricity balancing costs from Elexon BMRS half-hourly data. LEAR baseline vs XGBoost and LSTM.
PythonTime-seriesForecastingElexon BMRS
🦷
Dr Nastaran - Dental Website
Production site for a dental practice on Cloudflare Pages. GA4, GTM, Clarity, Web3Forms, before/after sliders, dark mode, mobile-first.
HTML/CSSJavaScriptCloudflare
🗄️
Bristol Air Quality - Dual DB Analysis
MySQL relational and MongoDB document analysis on Bristol air quality dataset. MSc distinction (84).
MySQLMongoDBPython
demos

Playground

📊
Retail KPI Dashboard
Interactive Streamlit dashboard over the £10.6M retail dataset. Revenue by country, product, and time.
⏳ Coming soon - deploying to Streamlit Cloud
🎙️
STT Pipeline Demo
Upload audio, get transcription. Faster Whisper tiny model, CPU-optimised.
⏳ Coming soon - live on Raspberry Pi
BM Cost Forecast Dashboard
Forecasting GB electricity balancing costs per settlement period. Dissertation project, live results as models are trained.
⏳ Coming - dissertation January 2027
stack

Tools

Languages
Python · R · SQL
JavaScript · HTML/CSS
C++ · C#
Data & ML
Time-series forecasting · XGBoost
Pandas · NumPy · Scikit-learn
LLM fine-tuning: QLoRA · Hugging Face PEFT
Voice pipelines: Faster Whisper · spaCy NER · Piper TTS · Ollama
Model evaluation: WER/CER · latency benchmarking
R: tidyverse, ggplot2, statistical modelling
Learning: LangChain · RAG · SHAP · GitHub Actions · n8n · Clarity
Databases
MySQL · PostgreSQL
MongoDB
Deploy & Infra
Cloudflare Pages · Render · Vercel
Supabase · Resend
FastAPI · Streamlit
Docker
GA4 · GTM

MSc Data Science student at UWE Bristol. I build systems that go from raw data to production - pipelines, models, APIs, and web platforms. BSc Computer Engineering, ranked 1st in graduating class.

Currently working on an AI-powered phone reception system for clinics (UWE Enterprise Scholarship, £1,000), and forecasting GB Balancing Mechanism costs with machine learning as my dissertation. Both involve real data, real constraints, and real deadlines.

I also build and deploy complete web systems end-to-end, from first commit to production. Everything I ship is live and in real use.

1st
BSc rank
8+
Projects
£1k
Scholar

Let's work together

Open to internships, graduate roles, and research collaborations.