Perfex Data Technology — Kathmandu, Nepal
Applied intelligence for
systems that can't afford
to fail.
We engineer intelligent software for organizations that depend on precision — and we build Lumaple, an Incident Intelligence System designed to detect, diagnose, and resolve failure before it reaches your customers.
Our position
01Software fails quietly, then all at once — and most teams find out from a customer before they find out from their own tools. We think the systems responsible for noticing, understanding, and resolving failure deserve the same engineering rigor as the systems they protect, powered by applied AI rather than a dashboard with extra steps.
That standard applies to everything we build — from custom machine learning systems for clients to Lumaple, our own product. We are a small, senior engineering team based in Kathmandu, structured to operate as a direct extension of the teams we work with, wherever they are.
What we build
Capabilities
Applied AI & Machine Learning
Custom models, LLM-powered products, and AI automation designed around your data — not a generic API wrapper.
Data Engineering & Platforms
Pipelines and analytics infrastructure that turn raw data into something a model can trust.
Incident Intelligence
Monitoring, correlation, and AI-assisted root-cause analysis — the discipline behind Lumaple.
Cloud-Native Engineering
Scalable, observable infrastructure with CI/CD and security built in from day one.
Product & Systems Design
Discovery and interface design that keep AI-powered products legible, trustworthy, and genuinely usable under pressure.
Flagship product — 03
Lumaple
Not just alerts — full incident intelligence.
Lumaple is an AI-powered Incident Intelligence System built by Perfex Data Technology. It sits above your existing monitoring stack — Zabbix, Datadog, Grafana, ELK, PagerDuty, Slack, and more — clustering related alerts, signals, and customer reports into one prioritized incident, then tracing the causal chain back to root cause and guiding your team to resolution with AI-generated summaries and one-click actions in Slack.
- 01Signal & alert clusteringRelated alerts, syslog, SNMP traps, and customer reports are clustered into one incident — not fifty duplicate tickets.
- 02Incident Graph & root causeA causal timeline traces every event back to root cause, with blast-radius mapping showing how the failure spread.
- 03Member AgentsPersonal AI for every responder — tracks availability, auto-acknowledges, and escalates when someone's unreachable.
- 04Slack-native responseAI-structured notifications with severity, summary, and one-click Acknowledge, Investigate, and Escalate actions.
- 00:00Signal cluster formed — auth p99 latency + 2 enterprise reports
- 00:14AI links cluster to incident: Authentication Failures
- 00:41Incident Graph traces root cause: TLS certificate expiry
- 02:07Member Agent auto-acknowledges · on-call notified in Slack
- 04:12Resolved — postmortem drafted automatically
How we work
Approach
Discover
We study your systems, data, and constraints before proposing a single line of code.
Design
Architecture and interface decisions are made deliberately, trade-offs stated plainly.
Build
Iterative development, tight feedback loops, and tests that mean something.
Operate
We ship with observability built in and stay engaged long after launch day.
Where we work
Built in Kathmandu. Engineered for the world.
Get in touch