AI and Machine Learning Implementation
The age of AI has arrived, and we are here to help
With over 8 years experience in the AI sector, we solve real-world problems by marrying research-grade statistical techniques with an agile, lean startup mentality.
Coefficient is recognized for rapidly delivering comprehensive AI-based products at a fixed price, all within a matter of weeks rather than months. Leveraging the power of AI, we design and deploy innovative solutions that fuel business expansion and prosperity.
Coefficient’s common services include…
Geospatial Modelling and Machine Learning
Natural Language Processing (NLP)
Recommender Systems
Open Source Predictive Modelling
Deep Learning for Object Detection
Company-wide Data Strategy and improving Data Maturity
Data Architecture and Predictive Analytics for Market Insights
CASE STUDIES
‘Technical Audit’ and Bias Mitigation in AI Driven Products
Winner of the Fairness Innovation Challenge, offering a unique socio-technical solution to address biases in AI-powered hiring products. By producing an open-source toolkit, conducting market evaluations and developing a 'technical audit' service, we aim to mitigate AI algorithmic bias in a wide range of use cases.
Predictive analytics for law enforcement
Delivered innovative tooling for law enforcement use by applying machine learning and geospatial modelling to live operational data. Deployment of web-based analytics tools using a Django-based API and React-powered front-end web application. Went from project kick-off to live deployment in an operational environment in 12 weeks.
Enhancing government policy decision-making for the FCDO with data analytics and PowerBI
Implemented two data analytics and modelling projects for the FCDO in 2023. Integrated various data sources from governmental statistical agencies, survey data and geospatial time series into clean analysis-ready data. Built predictive models and deployed operationally via PowerBI dashboards, enhancing the client’s situational awareness on the ground and supporting data-driven decision-making for the FCDO.
Multi-lingual NLP recommender system and data analytics strategy for health social network
Developed multi-lingual unsupervised NLP algorithms to detect related medical conditions and drugs for content on the HealthUnlocked website. Worked with the CEO to introduce and implement a company-wide data strategy centred on OKR-based practices, enhancing data-driven decision-making. Advised on data team strategy and led hiring round to recruit the company's first Data Science Lead.
Analysing supermarket shelves with deep learning
One-shot deep learning model for object detection using PyTorch for supermarket shelf analysis. Used Label Studio annotation framework to deliver dataset of annotated images.
Building early-warning systems for equity markets
In-house data team for parking space marketplace
We’ve built Franchise Partners’ data architecture from the ground up, automatically retrieving, validating and transforming a large number of “dirty datasets” with potentially market predictive properties for usage within both business intelligence and machine learning applications.
This included the Ingestion of financial ticker data plus alternative data sources into a data warehouse with automated ETL jobs. Machine learning R&D using a custom-built backtesting tool.
This allowed the team to Backtest stock selection models against historical investment opportunities.
Company-wide data strategy from KPI development to design of BigQuery data warehouse and rollout of Looker B.I. models and dashboards. Built geospatial machine learning models to forecast parking demand across the UK.