Who we are
Built from platform engineering.
Applied to AI systems.
LotusNex comes out of a background in production software engineering — not AI experimentation.
Over more than a decade building enterprise platforms, education systems, and process control software, our founder worked on systems where reliability, observability, and failure handling were not optional. These were environments where software had to behave predictably under real constraints.
When AI systems began entering production, a gap became clear.
The models were advancing quickly. The systems around them were not.
What failed was rarely the model itself — it was everything required to make that model usable in a real system: evaluation, boundaries, monitoring, ownership.
LotusNex was built to close that gap.
We apply the same engineering discipline used in production software — architecture, reliability, and operational clarity — to AI systems that are expected to run in the real world.
How we are structured
LotusNex operates as a focused engineering studio. Every engagement is led directly by the founder — architecture through handoff. Engagements are kept small by design: tight scope, clear interfaces, measurable outcomes.
What we bring
- Years of enterprise platform and cloud architecture delivery
- Deep DevSecOps and reliability engineering practice
- Production AI systems across agentic and retrieval domains
How we engage
- Architecture review before any build commitment
- Explicit evaluation criteria defined upfront
- Hardening, observability, and guardrails before handoff
- Clean documentation and operational ownership on exit
Who you'll work with
Om Thapa
Founder & Lead Engineer
Om has been building production software since 2011 across enterprise platforms, education systems, and process control environments — where reliability, system boundaries, and operational clarity are critical.
His background spans cloud architecture, reliability engineering, and large-scale systems integration, with a focus on systems that must perform under real-world constraints.
He founded LotusNex after consistently seeing AI systems break in production — not at the model layer, but in the surrounding architecture required to make them usable and reliable.
He works with teams that treat AI as production infrastructure — not experimentation.
Every LotusNex engagement is led directly by Om, from initial architecture review through final handoff.
LinkedIn →Operating principles
We build systems, not experiments.
Every engagement is scoped around production constraints: real users, real data, real operational burden. If a system cannot be evaluated, monitored, and owned, it is not ready to ship.
We take on fewer, more serious engagements.
LotusNex is selective by design. We engage where there is a real architectural problem and clear intent to ship production systems. We do not optimize for volume.
We build for the team that inherits the system.
Clean handoff is part of the work. Runbooks, interface documentation, observability baselines, and ownership clarity are not optional. The measure of good systems work is what happens after we leave.
What we optimize for
Technical rigor
- Clear system boundaries and explicit data contracts
- Evaluation-driven quality, not intuition-driven
- Deterministic fallbacks and well-defined failure modes
Operational readiness
- Telemetry, alerting, and cost visibility from day one
- Least-privilege access and auditability
- Runbooks and escalation paths before go-live
Long-term ownership
- Systems designed to be maintained, not worked around
- Documentation that reflects actual system behavior
- Interfaces internal teams can reason about
Client relationship
- Peer-level technical engagement, not vendor positioning
- Transparent scope and explicit constraints
- No work we cannot stand behind in production
If this matches how you think about building — let's talk →