AI
Applied AI, RAG & LLM Systems
AI features that impress in a demo often fall apart in production. I build LLM systems with proper retrieval, evaluation, guardrails and cost controls, so answers are grounded, measurable and affordable at scale.
Core
RAGLLMsFine-tuningGen AIClaude
Also in the toolbox
Anthropic ClaudeOpenAILlama / MistralpgvectorPineconeLangChainLlamaIndexHugging FaceEvals
What you get
Deliverables, not just hours.
01Retrieval-augmented knowledge assistants
02Document ingestion, chunking & embeddings
03Prompt, model and cost optimisation
04Fine-tuning on your domain data
05Evaluation suites and quality dashboards
06Guardrails, PII handling and audit logs
Typical use cases
Where this creates value.
Knowledge assistant
Ask questions across internal docs and get cited, permission-aware answers.
Document intelligence
Extract, classify and summarise contracts, invoices or claims automatically.
Domain fine-tuning
Smaller, cheaper models tuned to your vocabulary and output format.
Related work
AI in practice.
Explore other services