Open to Data Engineering, Data Science & ML roles · London

Junayd Elahi

I build data pipelines and machine learning projects, mostly with healthcare data. I care about clean data, work other people can rerun, and being honest about what a model can and can't do.

Now Data Analyst Coldwell Banker · I own the team's automation and data workflows
Degree BSc Computer Science (AI) University of Greenwich · 2:1
Focus Healthcare data & ML NHS, clinical & biomedical data
Manual work automated 0% of the weekly workload at Coldwell Banker

01 — Selected work

Projects

Each card is the short version. Open one for the full write-up: the method, the decisions, and what didn't work.

NHSBSA · 24 months
40GP practices above 1.5× the national rate, every month
2026·Personal project·Healthcare analytics

NHS Antibiotic Prescribing Outlier Detection

A Python and DuckDB pipeline over two years of public NHS prescribing data, finding GP practices that prescribe antibiotics far above the national rate. The rate is adjusted for seasonality, and the pipeline copes with the source changing format between releases.

Outcome 40 practices stayed above 1.5× the national rate in all 24 months, after excluding care homes, tiny lists and non-GP prescribers.

  • Python
  • DuckDB
  • SQL
  • Matplotlib
Read the deep dive Code ↗
Snowflake · dbt
39data-quality tests gate every build
2026·Data engineering

Healthcare ELT Pipeline

~160K rows of synthetic patient data loaded into Snowflake and modelled with dbt into a star schema. Orchestrated with Airflow in Docker, with CI on every pull request.

  • dbt
  • Snowflake
  • Airflow
  • Docker
Read the deep dive Code ↗
Final-year project
131participants · 51 timing features
2026·Dissertation

Keystroke Dynamics to Detect Parkinson's

Tests whether typing rhythm can separate Parkinson's patients from controls once data leakage is removed with subject-wise cross-validation.

  • XGBoost
  • Random Forest
  • scikit-learn
Read the deep dive Code ↗
PyTorch · CNN
0.98recall on the test set
2026·Deep learning

Pneumonia Detection from Chest X-Rays

A PyTorch CNN that classifies chest X-rays as normal or pneumonia, benchmarked against logistic regression, with its false positives and overfitting analysed.

  • PyTorch
  • CNN
  • Computer vision
Read the deep dive Code ↗
Go · PostgreSQL
50k/sevent bursts acknowledged in load tests
2026·Backend challenge

Sensor Streaming Backend

A Go service for room sensors. It ingests events, tracks device health and occupancy, and publishes fall alarms that consumers can resume after a disconnect.

  • Go
  • PostgreSQL
  • Docker
Read the deep dive Code ↗
Microscopy · BBBC008
19.7ppmean error vs expert masks
2026·Image analysis

Cell Coverage Estimation

Estimates how much of a fluorescence microscopy image is covered by cells using Otsu thresholding, then checks the answer against manually corrected reference masks.

  • Python
  • Otsu thresholding
  • Evaluation
Read the deep dive Code ↗

02 — About

Hi, I'm Junayd.

I'm a data analyst in London. I joined Coldwell Banker in 2023 doing data entry, got tired of typing the same things in every week, and wrote a script to do it for me. I showed it to my manager, and now I own the back-end data workflows and automation for our 4–5-person team: the pipelines, recurring processes and internal tools they rely on every day.

I finished my Computer Science (AI) degree at Greenwich in 2026. Most of my projects use healthcare data, because that's the area I'm most interested in. Some of my favourite results are the ones that didn't go to plan, like a model that stopped beating the baseline once I fixed a data leak.

Outside work

Most weeks I'm at the gym, on a bouldering wall or on a padel court, and I'm always up for a game of football or chess. At the weekend I'm usually working on my car, out driving or on my racing sim. I also like playing around with Arduinos and small electronics projects.

  • Gym
  • Bouldering & climbing
  • Padel
  • Football
  • Chess
  • Working on my car
  • Sim racing & driving
  • Arduino & electronics
Education
BSc Computer Science (AI), University of Greenwich, 2026 · 2:1 (69%)
Top modules
Statistics with R 86% · Advanced Mathematics 76% · Artificial Intelligence 74% · Natural Computing 73% · Machine Learning 70%
Languages
Python, SQL, Go, R

03 — Experience

Where I've worked & studied

  1. Oct 2023 — Present

    Data Analyst · Coldwell Banker, Southbank

    Started in data entry, automated recurring tasks, and now own the team's back-end data workflows, automation and internal tools supporting 4–5 colleagues.

    • Built Python automation tools that replaced recurring manual data-entry workflows, cutting weekly manual workload by 40%.
    • Scraped, cleaned and analysed sold-property data across 50 postcodes (around 2,500 listings) to produce local price trends for marketing and valuations.
    • Built an OCR-based matching pipeline that compares live listings against 270+ units using floorplan extraction and statistical scoring. It picks the correct unit about 94% of the time.
    • Automated a daily ETL pipeline on GitHub Actions. When runs started firing up to 10 hours late, I rebuilt the scheduling on a Cloudflare Worker, and it now starts on time every day.
  2. Oct 2025

    Quality Assurance Placement · Gnomen Property Management

    • Ran exploratory and boundary testing on pre-release UI, finding validation failures, race conditions and edge-case defects.
    • Reported defects with reproduction steps and severity, then presented findings and recommendations to the owner in non-technical terms.
  3. Sep 2023 — Jul 2026

    BSc Computer Science (AI) · University of Greenwich

    Upper Second-Class Honours (2:1), 69%. Final-year project on keystroke dynamics for Parkinson's detection.

04 — References

What managers have said

Quoted from written references. Full letters are available on request.

“It was entirely his own initiative that created the role he went on to hold. … He took these from idea to working tool largely independently.”
Bernadette TeumaDirector, Coldwell Banker Southbank
“He picked things up at pace and became productive in a short space of time. … often spotting things from the user-facing side that the development team did not have the same visibility of.”
Hassan N HassanManaging Director, Gnomen

05 — Toolkit

What I work with

AData engineering

  • dbt · Snowflake · DuckDB
  • Airflow · Docker · Kubernetes
  • GitHub Actions · CI/CD · Git
  • AWS S3 · CloudFront · Cloudflare Workers

BLanguages & analysis

  • Python · SQL · Go · R
  • Pandas · NumPy · statsmodels
  • Matplotlib · Power BI · Excel
  • PostgreSQL

CMachine learning

  • scikit-learn · XGBoost · Random Forest
  • PyTorch · TensorFlow
  • Nested, subject-wise cross-validation
  • Computer vision (CNNs, segmentation)

Certifications

  • Databricks Fundamentals AccreditationDatabricks Academy · Sep 2026
  • Software Engineer InternHackerRank role certification · Sep 2026
  • Learn Kubernetes · Learn GoBoot.dev · Jul 2026
  • Learn Power BI · Learn DockerBoot.dev · Jul 2026
  • Learn AWS · Learn SQLBoot.dev · Jun–Jul 2026
  • Python · Linux · GitBoot.dev · Sep 2025

06 — Contact

Let's talk.

I'm looking for Data Engineering, Data Science and Machine Learning roles in London. Healthcare is a bonus.