Translate ambiguity
I like problems where the technical, product, and organizational edges are still fuzzy. The work is to make them legible enough for a team to move.
About
I build AI systems by designing the experience around the agent.
I am Shashwat Jain, Head of Platform Engineering at mymediset in Singapore. I lead the company's shift from classical SAP extensions to cloud-native SaaS and build its vertical AI-agent platform for healthcare operations, in pilot with two enterprise healthcare customers. Outside work I build Ops Room, an operating agent for e-commerce founders, and run my own store to keep the work honest.
The professional spine
At mymediset I partner directly with the CTO on a healthcare SaaS platform used for medical-device loaner and consignment workflows — SAP BTP and SAP CAP, multi-tenant, in a regulated operational domain. Two shipped outcomes anchor that work: a mobile-first booking app (App Store · Google Play) that reduced medical instrument booking times by 50%, and an AWS object-detection system that reduced post-surgery kit identification effort by 90%. The current chapter is vertical AI agents for those same workflows — RAG, tool calling, and workflow orchestration — in pilot with two enterprise healthcare customers.
My work sits at the intersection of architecture, delivery, and leadership. I enjoy building systems where the hard part is not only the code, but also the boundaries: what the system is allowed to know, what it can do, how quality is measured, and how the team stays accountable as the product evolves.
AI has made that work more interesting. A good agent is not just a model with more tokens and more tools. It is a working environment with intent, context, evals, permissions, feedback, recovery, and human trust designed into it from the beginning.
User experience helps humans use products. Developer experience helps developers build with platforms. Agent experience is the next surface: the working environment that helps agents do useful work across support, operations, commerce, research, engineering, and administration.
LoopOps is my newest project in that direction. DevOps made delivery and operations continuous; LoopOps is my attempt to make improvement itself continuous, measurable, and agent-operable under human control.
Where I spend attention
- Agent experience for tool-using AI systems
- LoopOps, improvement infrastructure, and verified product loops
- Production AI architecture and enterprise platforms
- Evals, observability, permissions, and recovery paths
- Engineering leadership across ambiguous programs
- Commerce systems, Shopify ecosystems, and founder-led execution
How I lead
Design the environment
Agents create more value when the product gives them clear tools, state, permissions, examples, and recovery paths.
Keep judgment close to code
Architecture decisions are leadership decisions. I care about the tradeoffs, incentives, and failure modes behind the implementation.
Profiles
Project spotlight
Current projects that keep the work grounded
Commerce operations
Ops Room
An operating agent for e-commerce founders: a supervised, Cloudflare-native layer that watches the commerce stack, opens cases, grounds decisions in evidence, and executes approved changes through connectors — with an eval harness and nightly live runs in CI. Visit oproom.ai
Improvement infrastructure
LoopOps
Improvement infrastructure for agentic software teams: a way to turn telemetry, feedback, evals, issues, and verification windows into explicit loops that agents can operate and humans can approve. Visit loopops.devAgent-native hiring
HireWave
A job-kit utility for clearer public job pages. It turns role details into Markdown and JSON exports, schema-ready pages, clarity checks, and shareable URLs for candidates and agents. Visit hirewave.ai
Founder context