Applied AI · Backend systems · International remote engineering
Mohsin Hayat
Senior Software Engineer specializing in Applied AI and Backend Systems
I build production AI and backend systems where reliable engineering matters: conversational experiences, data-intensive agents, and the services around them. I’m currently at OnService.AI, building Python services and conversational AI experiences for airline workflows.

“One of those hires you notice from day one. Adapted incredibly fast, brought fresh technical perspectives, and consistently went beyond what was required.”
“Over this time he consistently proved his value as a smart, capable, autonomous and reliable software engineer.”
“Mohsin is a thoughtful, diligent, hard-working, and methodical engineer — a strong “slow thinker” (in a Kahneman/Tversky way).”
“Mohsin is a great person, a passionate developer and a hard working team member. He is not afraid of taking responsibility in difficult situations and always tries to raise the bar for himself.”
Selected work
Systems built for real operating contexts
SnippetGraph
- Problem
- AI teams need governed, dependable context instead of unversioned knowledge scattered across tools.
- Built
- A versioned knowledge-base authoring and publishing platform that composes snippets into pages and publishes immutable, hash-verified Git release artifacts through the GitHub GraphQL API.
- Context
- Designed for governed context publishing for AI-agent workflows.
- GitHub GraphQL API
- versioned content
- release artifacts
FirstCustomer
- Problem
- Product teams need a practical way to connect pull-request changes with the user flows they could affect.
- Built
- A CI-native browser agent that maps pull-request changes to affected user flows, tests real product paths with personas, and returns a confidence-scored release-risk assessment.
- Context
- Connects code changes to user-facing verification before release.
- CI
- browser agents
- release risk assessment
Production AI and backend systems
- Problem
- Operational AI needs dependable services, data boundaries, and understandable workflows—not just a model call.
- Built
- Conversational airline experiences, a production NL-to-SQL agent, a hierarchical budget copilot, and event-driven healthcare systems.
- Context
- Includes systems operating across 30+ tables, approximately $5M/month of planning context, and 100K+ monthly workflows.
- Python
- LangGraph
- KIU PSS
- AWS SNS/SQS
- ClickHouse
- CQRS
Capabilities
Focused engineering capabilities
- Applied AI and LLM agents
- Python and backend services
- Multi-agent orchestration
- NL-to-SQL and RAG
- Distributed and event-driven systems
- Data pipelines and analytical systems
- Production reliability and developer tooling
Public writing
Engineering notes and career retrospectives
The published posts are personal writing from 2022 on remote work and engineering. They are retained as dated perspectives, not presented as current AI case studies.
Let’s connect
Build reliable systems together
Based in Lahore, Pakistan, with 6+ years of international remote collaboration experience.