GenericML

GenericML is “human-in-the-loop AutoML” built as an application platform so domain experts can create, test, govern, and deploy models themselves, with full lineage in a knowledge graph. Most ML fails because it lives in notebooks; GenericML ships ML as a governed, domain-driven product workflow, UI-first experimentation, repeatable pipelines, and deployable endpoints, not ad-hoc scripts. GenericML has:- UI-driven ML for domain experts: choose data, features, targets, validation, retrain cadence. DDD-based model boundaries: models align to real subdomains (appointments, billing, clinical), reducing leakage and making ownership clear. Knowledge-graph governance: provenance, versioning, explainability, “why this prediction”, audit trails, and GraphRAG on your org’s data. Hence GenericML is a model factory + governance layer: it standardises how organisations build and operationalise many small models across many subdomains, with consistent deployment, monitoring, and re-training — while keeping humans in control. GenericML doesn’t replace domain math, it blends it. You can ensemble ML predictions with rules, constraints, optimisers, and simple statistical models, with explicit weights and change control. Hence GenericML turns machine learning into something domain teams can actually use. Instead of data scientists building one-off models in Python, we provide a governed ‘Model Lab’ where business users pick their data, targets, and constraints, run AutoML, compare candidates, ensemble models, and deploy safely — with full traceability stored in a knowledge graph. It’s ML as a repeatable product workflow, not a research project.

Any company that is sick of the cost and time it takes to build ML/AI models that then don't work. Human in the loop is required for model so that models can be validated by the domain experts before they reach production! 60-80 % of ML projects fail as there is no Human in the loop control and 95 % of Gen AI projects fail as they are not using knowledge graphs. With GenericML all applications are built with Domain Driven Design first as a templated approach using DDD, problems in each sub-domain, data vectors, mapping of data vectors to problems, models that can be generated, ensemble of models that could be used, types of ensembles and knowledge graph design and example queries.