PERTH-BASED AI ENGINEERING
We turn AI prototypes into production systems.
Libra helps enterprises establish the foundations, platforms and controls needed to adopt AI with confidence—then engineers the path from prototype to secure, observable production.
Perth, Western Australia · Mining, oil & gas and resources context · Founder-led engineering
Representative example
Production-readiness review
Identity & access
Roles, permissions and service boundaries
Evaluation gates
Quality baselines and release criteria
Monitoring & incident response
Signals, ownership and escalation paths
Release & rollback
Controlled deployment and recovery
Illustrative scope only. Not client work or evidence that a review occurred.
Services
Six ways we help enterprise AI move forward.
AI Enablement
Establish the architecture, tooling, operating model and team capability needed to adopt AI effectively.
Typical output: Foundation map, target architecture and adoption roadmap.
AI Prototype to Production
Turn an existing PoC, demo or prototype into a secure, scalable and monitored production system.
Typical output: Production-readiness plan, deployment path and operating controls.
AI Platform Engineering
Build the cloud infrastructure, model access, RAG patterns, identity, observability and deployment paths your teams can reuse.
Typical output: Reference platform, reusable patterns and engineering guardrails.
Agentic Workflow Development
Build multi-step AI workflows integrated with enterprise systems, approvals and exception paths.
Typical output: Workflow design, system integrations and human-control points.
LLM Evaluation Frameworks
Measure quality and reliability with evals, benchmarks, regression tests, guardrails and failure analysis.
Typical output: Evaluation suite, quality baselines and release gates.
AWS AI Consulting
Design and implement AWS-centric AI systems using services such as Bedrock, SageMaker and cloud-native infrastructure.
Typical output: AWS architecture, implementation plan and production foundation.
Representative example, not a completed-client claim
What this can look like in practice.
Engineering knowledge assistant
Permission-aware retrieval across controlled engineering documents, with source citations, specialist review and monitoring. This is a representative capability, not a claim of completed work.
- Source
- Retrieve
- Evaluate
- Human decision
Why Libra
Senior engineering, close to the work.
- Perth-based
- Local context and direct collaboration for Western Australian organisations.
- Founder-led delivery
- Work stays close to founding engineers, keeping decisions and feedback loops short.
- Platformisation and governance
- Reusable patterns, identity, evaluation, observability and controls from the start.
- Resources context
- Mining and oil & gas domain understanding shapes the data, integration and operating assumptions.
How we work
A direct path from question to operation.
1
Assess
Clarify the problem, users, systems, risk and evidence.
Problem frame and readiness findings.
2
Architect
Agree the target design, controls and production path.
Target architecture and delivery plan.
3
Build
Deliver in small, testable increments with real integration constraints.
Working increments and evaluation evidence.
4
Operate
Establish evaluation, monitoring, ownership and improvement loops.
Operating controls, runbooks and ownership.
Mahek Gupta, Founder, Libra Technologies
Founder-led, engineering-first.
Mahek Gupta is the founder of Libra Technologies.
View Mahek on LinkedInBring us the prototype, platform or workflow that needs to become dependable.
Contact is direct through LinkedIn. Nothing is booked automatically, and this site does not collect form data.