← Adam Tamer Get in touch
ElektaAug 2025 – Present
Experience

Putting AI to work where the work happens.

Elekta builds the machines that treat cancer with radiotherapy. I build the software that helps the people who make them.

I automate supplier quality and procurement work end to end, and prototype AI that turns years of non-conformance history into answers.

Part approval automationAI search over past fixesSupplier riskMicrosoft 365 automation
Loading 3D model

Elekta Harmony · 3D model by Elekta, via Sketchfab

800+parts fully automated through part-approval (PPAP) preparation this year
25minof admin time saved on every part
330h+of admin handed back to the team this year
10+tools, automations and prototypes across quality, procurement and AI
At a glance

Four areas, one pattern.

Find a process people repeat by hand, build the tool that does it properly, and put it in their hands. Everything below follows that pattern.

01

PPAP & quality automation

Part approvals prepared, reviewed and logged with almost no admin.

Jump to it ↗ 02

AI & predictive maintenance

Matching new problems to past non-conformances and their fixes.

Jump to it ↗ 03

Supplier risk & market intelligence

Commodity and index movements turned into supplier risk scores.

Jump to it ↗ 04

Process automation

Microsoft 365 and business-system workflows that take supplier admin off people's plates.

Jump to it ↗
01 · PPAP & quality automation

800+ parts approved without the paperwork.

Before a supplier's part can be used, it goes through PPAP (Production Part Approval Process). I automated the preparation end to end, so engineers review the evidence instead of assembling it.

  1. Flag the critical partsParts that need approval are identified automatically.
  2. Gather the documentsEngineering documents and the PSW (Part Submission Warrant) are pulled and prepared.
  3. File them properlyEverything lands in an organised supplier folder, structured by PPAP number, supplier, part number and purchase order (PO).
  4. Expert reviewThe supplier quality engineer (SDQ) compares the supplier's documents with our specification.
  5. Approve and updateOn approval, our systems are updated automatically, including delivery data such as weights and dimensions.
Elekta Versa HD radiotherapy machine
800+ parts fully automated this year · 25 min saved per part · 330+ h of admin removed
In use

PPAP preparation automation

The flow above: flagging, document and PSW preparation, folder structure, and automatic system updates on approval.

AutomationDocument handlingERP updates
In use

PPAP register status logic

Works out the true status of every record from part, supplier and revision history: full or interim approval, latest preparation, cancelled, duplicate requests, old revisions and new PPAPs. Applied across a large register.

Advanced ExcelData logic
Designed

PPAP scope portal

A replacement for the spreadsheet process: a form request creates the folders and documents, then tracks each PPAP through "SDQ needs scope", "With supplier" and "Awaiting review", with supplier lookup, revision history, AI-assisted sizing and OCR document review.

Microsoft FormsSharePointPower AppsOCR
02 · AI & predictive maintenance

Describe the problem. See every time it happened before.

Someone on the factory floor types a problem with a part in their own words. The assistant finds past cases where that part failed in a similar way, and shows what fixed it.

  1. Turn text into meaningEach non-conformance becomes an embedding: a list of numbers that captures what the problem is about, not the exact words used.
  2. Find the closest matchesA new problem is compared with the history using cosine similarity, so the same fault in different wording still matches.
  3. Group the patternsSimilar failures cluster together, showing the problems that keep coming back and the fixes that worked.
Elekta Infinity radiotherapy machine
Proof of concept

Failure clustering assistant

Embeddings and cosine similarity over historical non-conformance cases, finding clusters of similar failures and the fixes recorded against them.

EmbeddingsCosine similarityClustering
Research

Predictive maintenance data expansion

Connecting failure → repair → replacement part → machine behaviour, so quality, service and spare-parts data become an extra signal for machine predictive maintenance.

Data linkingPredictive maintenance
Concept

AI case routing

Agents that route each non-conformance to the right SDQ or supplier path, using AI where fixed rules break down instead of rigid rule-based RPA.

AI agentsWorkflow design

Builds on my dissertation on predictive maintenance for jet engines.

Under the hood

Years of records in.
One answer out.

Every line is a record from the build history. The ones that match light up and feed the answer.

03 · Supplier risk & market intelligence

See supplier risk before it lands.

A market intelligence web app that watches commodities, indexes and major market moves, then scores the risk they pose to each supplier.

  1. Watch the marketsCommodity prices and indexes are monitored, and major movements are flagged.
  2. Link moves to suppliersEach movement is connected to the materials and suppliers it affects, and what it could do to cost.
  3. Score the riskSuppliers get a risk score, so teams know where to act first.
Built

Market intelligence web app

Monitors commodities, indexes and major movements, and assesses supplier risk with risk scores.

Web appRisk scoringMarket data
In use

Supplier risk mitigation tool

Standardised how SDQs identify, record and track supplier risk, so everyone works from one consistent method.

Advanced ExcelRisk logic
Built

Commodity & index mover report

Reports which commodities and indexes moved, and links each change to supplier and material cost exposure.

Power QueryDashboards
Built

Supplier & PPAP dashboards

Large operational datasets turned into dashboards that procurement and supplier teams can act on.

Power QueryData modellingReporting
04 · Process automation

Taking the repetitive work off people's plates.

Supplier paperwork used to live in inboxes and spreadsheets. Microsoft 365 workflows now capture it, file it and update the business system (ERP), the same way every time.

  1. Capture the requestA form or a flagged record starts the flow, instead of an email someone has to remember.
  2. File it properlySupplier documents land in the right SharePoint folder, named consistently, so they can be found later.
  3. Update the systemsRegisters and the business system (Infor M3) are updated by the flow, so nobody re-keys the same data.

Individually less flashy, together they show the habit: spot the manual steps, build the tool, hand it over.

Built

Microsoft 365 supplier workflows

Flows for ISO certificate requests, PSW preparation and supplier documentation, with documents filed to SharePoint automatically.

Power AutomateSharePointMicrosoft Forms
Built

Supplier and ERP admin automation

Cut the repetitive administration around supplier cases and information handling in the Infor M3 ERP.

Infor M3Process automation
Built

Supplier admin toolkit

Smaller tools for registers, folders, reporting and data preparation across the supplier team.

Advanced ExcelPower Query
Skills

What I bring to a team.

Grouped for a quick read: the techniques, the tools and the domain knowledge behind the work above.

AI & data

  • Text embeddings and vectorisation
  • Cosine similarity search
  • Failure clustering
  • Predictive maintenance
  • AI agents for workflow routing
  • OCR for document review

Automation & tools

  • Power Automate
  • Advanced Excel logic at scale
  • Power Query and dashboards
  • Microsoft Forms, SharePoint, Power Apps
  • Web apps

Domain

  • PPAP and PSW
  • Supplier quality (SDQ) workflows
  • Non-conformance management
  • Procurement and quotations
  • Commodity and index pricing
  • Infor M3 ERP

How I work

  • Map the process before automating it
  • Build tools people actually adopt
  • Measure the time saved
  • Choose AI or rules on merit
  • Prototype fast, then harden