Service
AI and ML product design: MLOps platforms, assistants and generative tools
I've designed an MLOps platform (Orchestra), a generative marketing tool (Assemble) and an ERP with an AI assistant inside it (369 ai ERP). The hard part is rarely the model. It's making the system legible to the people running it.
Who it is for
- AI startups turning a model or API into a product people can use
- Platform teams building internal ML tooling
- Enterprise products adding an AI assistant to existing workflows
I'm also available as a white-label designer for agencies that outsource design to India.
What's included
- Flows for starting, configuring and governing workloads (compute, GPUs, environments)
- Monitoring views that show what is running, what failed and why
- AI assistants placed inside the workflow, with clear inputs and reviewable outputs
- Generation flows with brand or policy controls
- Empty, loading, long-running and error states designed up front
- Developer-ready specs for complex tables, logs and charts
Related work
- Orchestra: MLOps: GPUs, Kubernetes, monitoring, LLM inference
- Assemble: Generates on-brand landing pages and emails
- 369 ai ERP: Multi-tenant ERP, 24 modules, in-context AI assistant
Ways to work together
Fixed-scope project
A defined piece of work, such as a new product area, a redesign or a deck, with an agreed scope and timeline.
Monthly retainer
A set amount of design time each month for ongoing product work and design-system upkeep.
Dedicated designer (staff augmentation)
I join your team for a fixed period and work in your tools, rituals and sprint cycle.
White-label for agencies
I design under your agency's name. The client relationship stays with you.
How it runs
- BriefYou send what you have: the product, the users, constraints and deadlines. We talk it through on a call.
- UnderstandResearch and journey mapping with your users and stakeholders, and a review of what exists today.
- StructureInformation architecture, flows and wireframes, agreed before any visual polish.
- Design and testUI and interaction design, a clickable prototype, and usability testing where we can get users.
- HandoffSpecs with states, edge cases and system guidance. I stay available while engineering builds.
Time zones
I work from Thiruvananthapuram on India Standard Time (IST, UTC+5:30). These are the overlap windows I can offer for calls and reviews:
| Your team | Your time | My time (IST) |
|---|---|---|
| UK (GMT) | 09:00 – 13:00 | 14:30 – 18:30 |
| Central Europe (CET) | 09:00 – 14:00 | 13:30 – 18:30 |
| Gulf (GST) | 08:00 – 17:00 | 09:30 – 18:30 |
| US East (ET) | 08:00 – 10:00 | 18:30 – 20:30 |
| Australia East (AEDT) | 14:00 – 18:00 | 08:30 – 12:30 |
Indicative. Daylight saving moves these by an hour. Most of the work itself is asynchronous.
Questions
Have you designed AI products before?
Yes: Orchestra (MLOps: GPU allocation, Kubernetes orchestration, live monitoring and LLM inference), Assemble (AI-generated landing pages and emails) and 369 ai ERP (an in-context AI assistant across 24 modules).
Do we need a working model before design starts?
No. Design can start from your capabilities and constraints, and adjust as the model and API settle.
Can you design for both technical and business users?
That is most of the job on platforms like these. I usually map who does what first, then decide how much of the system each role needs to see.
Got a product that needs a senior designer?
Send a short note about the product, the team and the timing. Email is the quickest way to reach me.