>>>>>>> TestZone =======
>>>>>>> TestZone

Is This Your Situation?

Good fit
  • Evaluating a vendor's AI proposal
  • A previous AI project didn't deliver
  • Tools your team isn't actually using
  • Manual process AI could probably fix
  • Need something built end-to-end
Not the right fit
  • Regulatory or legal AI compliance
  • Enterprise governance programmes
  • Academic or research AI work

Two Ways I Can Help

Advisory calls are for businesses that need a clear second opinion before spending. Is this vendor's quote realistic? Will this tool work for your actual workflow? One conversation can save weeks of the wrong direction.

Build engagements are for businesses with a defined problem that needs solving. I build LLM integrations, RAG systems, and agentic workflows that replace manual processes — end to end, in production.

Book a free advisory call
Discuss a build project →

Recent Work

I'm currently the sole AI developer at a B2B software company, shipping production systems that real users depend on.

4 hrs → minutes
Manual review eliminated

Automated scientific article analysis pipeline: keyword categorisation, LLM relevance scoring, and configurable routing. Batch processes 200 articles — auto-include, human review, or discard.

97.5%
Field completion rate

Contact data enrichment engine using Google Search API and GPT-4o, with context-aware validation — date verification, source reliability, URL credibility scoring.

10 tool endpoints
Agentic workflow assistant

AWS Bedrock Agents system letting users build and modify enterprise approval workflows through natural language. Full-stack: React, Node.js backend, multi-environment AWS deployment.

What an Advisory Call Looks Like

You describe your situation. I tell you what I'd actually do — including whether AI is the right answer at all.

It's a direct conversation with someone building these systems right now, not a consultant working from a slide deck. Topics I've covered: ComfyUI workflow optimisation, evaluating AI vendor proposals, RAG system design, why a chatbot isn't performing, build vs. buy decisions.

The first call is free. If there's more to dig into, we go from there.

Book a free call
Miles Johnson

Background

I'm an AI engineer based in Melbourne, currently the sole AI developer in the R&D division of a chemical management software company — architecting and shipping production AI systems from scratch.

I've been building software since 2019, with the last two years focused entirely on AI engineering: agentic tool-calling on AWS Bedrock, RAG pipelines, automated document processing, full-stack integrations. I also do freelance work including ComfyUI workflow optimisation and custom LoRA training for production studios.

Practical upshot: I know what these systems actually cost to build, where they break, and where vendors tend to oversell.

Technical Stack

AI & LLM

AWS Bedrock Agents, OpenAI API, Anthropic API, Gemini API, RAG systems, agentic tool-calling, OpenAPI schema design, prompt engineering, LLM evaluation, ComfyUI, LoRA training.

Backend & Cloud

Node.js, Python, Express, serverless architecture, AWS (Lambda, EC2, S3, Bedrock, Cognito, Lex), Docker, RESTful and WebSocket APIs.

Frontend & Data

React, TypeScript, Redux, MongoDB, PostgreSQL, Redis, Pandas, BeautifulSoup. Full-stack from architecture to deployed application.