Skip to content
Tillett Solutions

Analytics · Prediction models · Applied ML

Turn data into insight, predictions and useful tools.

Data analytics, prediction models and practical automation for problems in many different fields.

Explore services Selected projects

Areas of work

  • Prediction models
  • Data analysis
  • Machine learning
  • AI-assisted development
  • Automation
  • Data pipelines
  • Computer vision
  • Analyticstrends, segments and reporting that lead to decisions
  • Prediction modelsbuilt, tested and refined end to end
  • Tools & automationanalyses and models turned into tools used every day

Services

Analytics, prediction models and applied machine learning

Data & prediction

  • Prediction models

    Predictive modelling: build, test and refine useful models.

  • Data analysis & reporting

    Trends, period comparisons and clear reports that end in a decision.

  • Segmentation & diagnosis

    Find what changed and why across different groups.

  • Data pipelines & quality

    Reliable collection, cleaning and revision tracking.

Machine learning

  • Computer vision

    Detection and classification on images and video.

  • Hybrid vision + OCR

    Feature matching, OCR and rules alongside learned models.

  • Reinforcement learning

    Self-play, search and policy/value networks for games.

Engineering practice

  • Dataset engineering

    Collect, label and prepare data for reliable results.

  • AI-assisted development

    Agent teams that write, test and run data and ML code under review.

  • Automation & useful tools

    Turn analyses and models into everyday tools.

  • Evaluation & experimentation

    Test against baselines and check for regressions.

Approach

AI is the workshop, not just the product.

AI-assisted development, guided by human judgement.

Objectives & judgementhuman-led

  • Collect & clean data
  • Analyse it
  • Prediction models
  • Training code
  • Test & evaluate
  • Tools people use

Process

From data to working models

Analytics

From raw data to a decision

  1. 01Collect & clean
  2. 02Compare & segment
  3. 03Diagnose
  4. 04Recommend & measure

Model development

Building a model

  1. 01Data preparation
  2. 02Feature & model design
  3. 03Evaluation
  4. 04Iteration
  5. 05Deployment

Selected services

Services in practice

Workflow concept: search and on-site behaviour signals are combined, partial days are labelled, and findings become ranked actions Search On-site Combine final dayspartial Actions 1 2 3 with a metric

Data analysis & decision support

Bring data together, uncover patterns and turn findings into clear decisions.

Pipeline concept: price, indicator and search-trend series are collected, normalised, quality-checked with revisions tracked, then queried, with provenance kept throughout Prices Indicators Search trends CollectNormalise CheckQuery revisions Provenance kept at every step

Data & time-series pipelines

Collect, clean and organise dependable data, with revisions and gaps tracked.

Concept: a connected personal AI system: Data → Personal model → Action → Feedback → Improve → Data. C2, Chat and C2 Mini connect to the cycle. Action includes physical home actions. Feedback from outcomes and user corrections feeds improvement. C2 Chat C2 Mini Data Personal model Action Feedback Improve

Personal AI & adaptive automation

Connect assistants, conversations and everyday workflows that adapt to feedback.

Concept: generic playing cards with one hidden card, a branching search over possible plays, and a self-play, train, gate and promote loop ? hidden card search Self-play Train Gate Promote

Hidden-variable modelling

Explore hidden states and compare decisions through search and simulation.

Concept: a stylised book cover whose band regions feed band-presence and band-design heads; together with an identity embedding they pick the matching edition Band presence Band design Embedding matched edition

Multi-task photo identifier

Identify objects and recognise multiple visual attributes from a single photo.

Smart home

A personal AI system that keeps learning

A connected personal AI system for work, conversation and home, designed to improve through corrections and periodic retraining.

Personal AI learning cycle: Data → Personal model → Action → Feedback → Improve → Data C2, Chat and C2 Mini connect to the learning cycle, with You linked to Feedback. Data feeds a personal model, which informs action, including physical home actions. Feedback from outcomes and user corrections feeds improvement, then the cycle returns to data. C2 Chat C2 Mini You Data Personal model Action Feedback Improve Personal AI learning cycle: Data → Personal model → Action → Feedback → Improve → Data C2, Chat and C2 Mini connect to the learning cycle, with You linked to Feedback. Data feeds a personal model, which informs action, including physical home actions. Feedback from outcomes and user corrections feeds improvement, then the cycle returns to data. C2 Chat C2 Mini You Data Personal model Action Feedback Improve
  • C2Complex work & agents
  • C2 MiniEveryday requests
  • ChatConversation & workspace
  • Home integrationsPhysical home
  • Event & pattern modelsEvents & patterns
  • Personal LLMTailored conversation
  1. 1Data
  2. 2Personal model
  3. 3Action
  4. 4Feedback
  5. 5Improve
Portrait of me: head and shoulders, dark jacket, neutral background

About me

A decade of machine learning, now accelerated by AI

I develop models, analyse data and build practical tools for problems in many different fields: if there is data behind a question, I can usually help answer it.

I use AI assistants and agent teams to build faster, with human judgement guiding the work.

  • 10 yearsbuilding machine learning systems
  • Many fieldsvision, games, analytics, data platforms and smart home