M. ENGINEERING SYSTEMS
PORTFOLIO / 2026
INITIALIZING 00%
M.
A LITTLE SPACE TO BUILD BIG THINGS

WELCOME.

TO MOMNA'S ENGINEERING WORLD

CLOUD✳ DEVOPS✳ AUTOMATION✳ RELIABILITY✳ KUBERNETES✳ CLOUD✳ DEVOPS✳ AUTOMATION✳ RELIABILITY✳ KUBERNETES✳
SYSTEMS WITH PURPOSE
01 — 06
CASE STUDY 01 / OBSERVABILITY ENGINEERING
SAAS OBSERVABILITY

See every
signal. Securely.

Multi-Tenant SaaS Observability Architecture

Designing a unified observability platform that brings distributed traces, metrics and logs together while addressing tenant-level access control across a multi-cloud environment.

OPENTELEMETRY GRAFANA TEMPO PROMETHEUS CORTEX FLUENT BIT LOKI KEYCLOAK OPA TERRAFORM
i

Implementation status: unverified. The available source is a project presentation; independent deployment and tenant-isolation test evidence has not been established.

OBSERVABILITY PIPELINE 01 — 05
01
Tenant Applications Application telemetry sources
↘
02
OpenTelemetry Collect traces, metrics and logs
↘
03
Telemetry Backends Tempo · Cortex · Loki
↘
04
Grafana + Alertmanager Dashboards and metric-based alerts
↘
05
Tenant Access Controls Keycloak identity + OPA policies
SECURITY
TRACES / METRICS / LOGS AWS + IBM CLOUD
PROJECT TYPE SaaS Observability
TELEMETRY Traces / Metrics / Logs
SECURITY DESIGN Keycloak / OPA
CLOUD ENVIRONMENT AWS / IBM Cloud
ENGINEERING REQUIREMENTS OBS-01
R01

Unified telemetry collection

Define how application telemetry is collected and routed. The presentation identifies OpenTelemetry as the collection layer for distributed traces, metrics and logs.

OTEL / TELEMETRY
R02

Tenant-aware observability

Establish how each SaaS customer can inspect its own application signals. The design describes tenant-specific metrics and logs; enforcement still needs technical evidence.

MULTI-TENANCY
R03

Access control and isolation

Define tenant identity, authorization policies and controls that must prevent cross-tenant telemetry access. Keycloak and Open Policy Agent (OPA) are the proposed security components in the source architecture.

IDENTITY / AUTHORIZATION
R04

Investigation and alerting

Specify how dashboards bring metrics, traces and logs together, and how metric thresholds trigger alerts. The presentation identifies Grafana and Prometheus Alertmanager for this workflow.

DASHBOARDS / ALERTS
R05

Multi-cloud deployment design

Document the intended AWS and IBM Cloud environment and Terraform-based provisioning. Actual deployment, failover and recovery must be verified separately.

TERRAFORM / CLOUD
REQUIREMENTS TO VALIDATE R01 — R05
02 / PROBLEM & REQUIREMENTS
THE ENGINEERING CHALLENGE

Visibility without cross-tenant exposure.

A SaaS observability platform must make application behavior easier to investigate without exposing one customer's telemetry to another. Collecting signals is only part of the challenge: tenant context, access enforcement and trustworthy operational evidence are equally important.

01 PROJECT OBJECTIVE

Design a multi-tenant observability platform that combines traces, metrics and logs, supports dashboard-led investigation and metric-based alerting, and defines tenant-level access controls across an intended AWS and IBM Cloud environment.

ENGINEERING SCOPE
01 Collect → Process → Store telemetry
02 Visualize → Alert → Investigate
03 Authenticate → Authorize → Isolate tenants
04 Provision → Validate → Document evidence
!

Evidence boundary: The supplied 16-page presentation describes the architecture and calls the platform fully functional in its conclusion. However, the presentation alone does not verify a live deployment, configured alert rules, or successful cross-tenant isolation tests. These claims must remain unverified until supporting evidence is available.

03 / ARCHITECTURE & SYSTEM DESIGN

One signal path.
Separate tenants.

The proposed platform collects application telemetry through OpenTelemetry, routes each signal to a purpose-built backend, and brings operational investigation together through Grafana. Tenant identity and authorization form a separate security concern that must be enforced throughout the data-access path.

CORE TECHNOLOGIES
OpenTelemetry Grafana Tempo Prometheus Cortex Fluent Bit Loki Grafana Keycloak OPA Terraform
Tenant isolation model DESIGN REQUIREMENT
Tenant A Own identity, authorized telemetry queries and permitted views.
Tenant B Own identity, authorized telemetry queries and permitted views.

Required boundary: each request must be authorized for the correct tenant, and tenant context must be enforced at the relevant ingestion, storage and query boundaries—not merely shown as separate boxes in a diagram.

Telemetry architecture

Conceptual flow
01

Tenant applications

Application services produce telemetry for their workloads.

SOURCES
02

OpenTelemetry collection

Collect and route distributed traces, metrics and logs.

INGESTION
03

Signal-specific backends

Route each telemetry type to its corresponding storage/query service.

PROCESSING
04

Grafana dashboards

Correlate metrics, traces and logs during investigation.

VISUALIZATION
05

Metric-based alerting

Prometheus Alertmanager handles alerts from configured rules.

ALERTS

Telemetry backend mapping

THREE SIGNAL TYPES
Tempo Distributed traces
Prometheus + Cortex Metrics and scalable storage
Fluent Bit + Loki Log collection and aggregation

This mapping follows the tool stack described in the project presentation. Exact deployment configuration and runtime connections have not yet been independently verified.

SECURITY LAYER Keycloak + OPA

Identity management and policy-based authorization are described in the PDF.

INFRASTRUCTURE Terraform · AWS · IBM Cloud

Intended infrastructure provisioning and cross-cloud environment.

↳ Architecture status: documented design; deployment status unverified. The diagram alone does not prove tenant isolation, successful failover, or that alert rules have been configured and executed.

04 / IMPLEMENTATION & CORE COMPONENTS

Every signal has
a purpose.

The proposed observability stack separates telemetry collection, signal-specific storage, visualization, access control and infrastructure provisioning. Each component has a distinct responsibility in the end-to-end design.

ENGINEERING PRINCIPLE

Separate signal processing from identity and authorization. Keycloak provides the identity layer and OPA is intended to evaluate access policies, but tenant isolation depends on those policies being enforced throughout ingestion, storage and query paths. Actual configuration and test results are not established by the presentation alone.

COMPONENT / 01 Collection

OpenTelemetry

Collects application telemetry, including distributed traces, metrics and logs, for routing into the observability stack.

Role: Telemetry ingestion
COMPONENT / 02 Tracing

Grafana Tempo

Stores and queries distributed trace data to help investigate requests as they travel through application services.

Role: Trace storage / queries
COMPONENT / 03 Metrics

Prometheus + Cortex

Prometheus collects operational metrics such as CPU, memory and request latency. Cortex is identified as the scalable, multi-tenant metrics component.

Role: Metrics / scalable storage
COMPONENT / 04 Logging

Fluent Bit + Loki

Fluent Bit ships logs to Loki, which indexes metadata for log aggregation and tenant-specific diagnostic workflows.

Role: Log shipping / aggregation
COMPONENT / 05 Security

Keycloak + OPA

Keycloak handles identity and login; Open Policy Agent (OPA) is described as enforcing role-based and granular access policies for tenant-level restrictions.

Role: Authentication / authorization
COMPONENT / 06 Visualization

Grafana + Alertmanager

Grafana brings traces, metrics and logs together for investigation. Prometheus Alertmanager handles metric-based alerts triggered by configured thresholds.

Role: Dashboards / alert delivery
COMPONENT / 07 Infrastructure

Terraform + Cloud

Terraform is identified for infrastructure provisioning across the intended AWS and IBM Cloud environment. Failover and disaster recovery require separate implementation evidence.

Role: Infrastructure provisioning
05 / WORKFLOW & EXECUTION
TELEMETRY LIFECYCLE

From signal to insight.

The proposed workflow turns raw application signals into actionable operational context. Traces help follow a request, metrics reveal performance changes, and logs provide detailed events for investigation.

EXECUTION PRINCIPLE

A dashboard or alert is useful only when its underlying telemetry can be trusted and accessed by the right tenant. Collection, routing, authorization and query boundaries must work together; the presentation does not provide runtime evidence that these steps have been executed successfully.

SIGNAL-TO-BACKEND MAPPING

TRACES OpenTelemetry → Grafana Tempo
METRICS Prometheus → Cortex
LOGS Fluent Bit → Loki
OBSERVABILITY EXECUTION SEQUENCE 01 — 06
01

Tenant application activity

SOURCE

A request moves through application services—for example, a customer checkout journey. The services produce the operational events and signals to be observed.

Application → telemetry signals
02

Collect and route telemetry

INGESTION

OpenTelemetry is the collection layer described in the PDF. Telemetry must retain trustworthy tenant context as it moves toward the relevant processing and storage services.

OpenTelemetry / tenant context
03

Process and store each signal

BACKENDS

Traces are handled by Tempo, metrics by Prometheus and Cortex, and logs by Fluent Bit and Loki. Tenant-aware storage and queries must preserve the intended boundaries.

Tempo · Cortex · Loki
04

Authenticate and authorize access

SECURITY

Keycloak is described for tenant identity and login; OPA is described for access policies. The request must be checked against the tenant it is permitted to access.

Identity → policy decision → permitted access
05

Visualize and trigger alerts

OBSERVE

Grafana combines metrics, traces and logs for investigation. Configured Prometheus alert rules can detect thresholds; Alertmanager handles resulting alerts.

Dashboard / metric threshold / alert
06

Investigate the cause

DIAGNOSIS

The PDF describes an investigation sequence: a high-latency alert leads to a trace inspection, then related logs can help identify a slow database call or timeout.

Alert → trace → correlated logs
INTENDED OUTCOME

Tenant-aware operational investigation

Verification status: this is a workflow reconstructed from the project PDF, not a captured execution trace. Actual dashboards, alert rules, access-policy configuration and successful runs still require supporting evidence.

DOCUMENTED PROJECT ARTIFACTS

Evidence, not assumptions.

04 REVIEW AREAS
EVIDENCE / 01 DOCUMENTED IN PDF
01 Tenant apps
02 OpenTelemetry
03 Telemetry stores

System architecture & telemetry routing

The source diagram identifies the collection layer, trace/metric/log backends, security services and cross-cloud infrastructure.

SOURCE: PDF / PAGE 4
EVIDENCE / 02 STACK DESCRIBED
TRACE Tempo
METRIC Cortex
LOG Loki

Three telemetry signals

The PDF describes separate trace, metric and log paths using Tempo, Prometheus/Cortex and Fluent Bit/Loki.

SOURCE: PDF / PAGES 5–7
EVIDENCE / 03 WORKFLOW DESCRIBED
Alert High latency
Investigate Trace + logs

Dashboard & alert correlation

Page 9 illustrates the intended investigation path: latency alert, trace inspection and correlated logs. Actual rule configuration is not supplied here.

SOURCE: PDF / PAGE 9
EVIDENCE / 04 RUNTIME TEST NOT PROVIDED
Tenant A Own telemetry
Tenant B Own telemetry

ISOLATION INTENT — NOT A TEST RESULT

Tenant isolation needs independent proof

Keycloak identity and OPA policy enforcement are described on page 8. No executable policy, backend access-control configuration or cross-tenant denial test is included in the evidence shown here.

SOURCE: PDF / PAGE 8 · TEST EVIDENCE OUTSTANDING

Evidence boundary: The mini diagrams above are explanatory visuals created for this portfolio section, not screenshots from a running system. The PDF is the source document for the architecture and workflow descriptions.

VERIFY THE CLAIMS

Designed to observe. Evidence to verify.

The project PDF describes a multi-tenant observability platform across AWS and IBM Cloud. This review separates what the presentation documents from what would still need to be demonstrated through runtime artifacts and tests.

01

Telemetry flow

Pages 4–7 describe OpenTelemetry collection and the separate trace, metric and log backends. These pages document the design, not a captured successful ingestion run.

02

Dashboards & alert rules

Page 9 describes Grafana visualization and threshold-based alerting with Prometheus Alertmanager. Actual dashboard exports, rule definitions and alert-firing evidence have not been supplied.

03

Tenant-isolation validation

Page 8 names Keycloak and OPA as identity and authorization components. To verify isolation, evidence must show that Tenant A cannot retrieve Tenant B's traces, metrics or logs.

CURRENT EVIDENCE STATUS

Documented architecture; deployment status unverified. The supplied PDF is sufficient to explain the intended design, but not to establish that the system is deployed, that alerts have fired, or that cross-tenant access has been blocked in practice.

ENGINEERING THE TRUST BOUNDARY

Visibility without cross-tenant exposure.

A multi-tenant observability platform must protect more than application data. Traces, logs, metrics, dashboards and alerts can reveal customer behaviour and internal system details. Identity, authorization and tenant-aware data access therefore belong in the design from ingestion through investigation.

CORE SECURITY PRINCIPLE

Authenticate the user. Authorize every data request.

The source PDF identifies Keycloak for identity management and Open Policy Agent (OPA) for authorization. These components describe the intended control layer; effective isolation also depends on enforcing tenant context in the telemetry access path.

The isolation boundary

Tenant A Authorized data scope
Tenant B Separate data scope

DESIGN INTENT · CROSS-TENANT DENIAL STILL REQUIRES TEST EVIDENCE

RELIABILITY REVIEW 04 / CONTROL AREAS
01 / ACCESS
IDENTITY & POLICY

Tenant-aware authorization

Keycloak is identified for tenant identity and login; OPA is identified for policy-based authorization. The access layer must check the authenticated tenant against the requested data scope before returning telemetry.

Keycloak OPA RBAC
02 / ISOLATION
TELEMETRY ACCESS

Protect every signal

Tenant boundaries must hold across traces in Tempo, metrics in Prometheus/Cortex, logs in Loki, and their dashboards and alert views. Separate labels or a diagram alone do not establish secure isolation.

Tempo Cortex Loki
03 / DETECTION
ALERTING & DIAGNOSIS

Make failures observable

The PDF describes threshold-based alerts with Prometheus Alertmanager and investigation through Grafana, traces and logs. Actual thresholds, alert delivery and incident response have not been verified from the supplied document.

Prometheus Alertmanager Grafana
04 / RECOVERY
CLOUD RESILIENCE

Plan for service disruption

The PDF proposes Terraform-managed infrastructure across AWS and IBM Cloud, with failover, backups and disaster recovery. Automatic failover and successful recovery should only be claimed after deployment and recovery tests demonstrate them.

Terraform AWS IBM Cloud

Security & reliability verification boundary

Current status: design documented; runtime implementation unverified. The supplied PDF describes the intended controls, but does not include executable Keycloak/OPA configuration, cross-tenant denial results, live alert tests or successful failover evidence. These are required before claiming the controls work in practice.

ENGINEERING ANALYSIS / 08

Observability decision matrix

05 COMPONENTS
DOCUMENTED TELEMETRY FLOW
OpenTelemetry Collect traces, metrics and logs
TRACE
Grafana Tempo Distributed trace storage and queries
METRIC
Prometheus + Cortex Metrics collection and multi-tenant scaling
LOG
Fluent Bit + Loki Log forwarding and aggregation
IAM
TENANT ACCESS CONTROL

Keycloak handles tenant identity; Open Policy Agent (OPA) is intended to enforce granular authorization policies. Isolation must be verified at actual data-access boundaries.

TECHNOLOGY ENGINEERING VALUE TRADE-OFF
01 OpenTelemetry + Tempo

Distributed request tracing across services

Instrumentation, trace retention and query performance

02 Prometheus + Cortex

Metrics collection with a multi-tenant scaling approach

Tenant identity, query boundaries and capacity planning

03 Fluent Bit + Loki

Centralized logs with metadata-based exploration

Correct tenant labels, access policies and log retention

04 Keycloak + OPA

Identity management and policy-based authorization

Policies must be enforced and tested across access paths

05 Terraform + AWS / IBM Cloud

Infrastructure provisioning across cloud providers

Configuration drift, networking and recovery complexity

∴
DESIGN IMPLICATION

Multiple backends improve separation of responsibilities. Reliable cross-signal investigation and tenant isolation, however, depend on consistent metadata, correct configuration and verified authorization—not merely the selected tools.

SYSTEM DESIGN TRADE-OFFS

One platform. Three signals. Clear boundaries.

The supplied architecture presentation describes an open-source observability stack that separates traces, metrics and logs into purpose-specific backends. The design combines these signals for investigation while introducing tenant-aware identity and access control across a proposed multi-cloud environment.

A

Choose the right backend for each signal

Tempo handles traces, Prometheus/Cortex handles metrics, and Fluent Bit/Loki handles logs. Correlation requires compatible metadata and working integrations.

B

Make tenant boundaries explicit

The presentation assigns identity management to Keycloak and policy-based authorization to OPA. To prevent cross-tenant exposure, tenant context must be preserved and enforced at ingestion, storage and query boundaries.

C

Balance cloud flexibility with operations

Terraform and AWS/IBM Cloud form the documented infrastructure direction. A multi-cloud design can support flexibility, but failover, backups and high availability need implementation and recovery evidence.

i
IMPLEMENTATION STATUS

The supplied PDF describes the platform as fully functional and presents an intended architecture. However, the PDF alone does not independently verify a live deployment, effective tenant isolation, or successful cross-tenant denial tests. These claims remain unverified from the available evidence.

09 / RESULTS & REFLECTION ARCHITECTURE REVIEW

What the platform is designed to demonstrate.

The supplied presentation describes an observability platform combining traces, metrics and logs with tenant-aware access controls. Its intended value is clearer troubleshooting and more isolated customer monitoring—not a performance result that has already been independently measured.

∴

Engineering takeaway An architecture can explain how isolation should work. Evidence from real policies, configuration and negative access tests is needed to show that isolation actually works.

OPENTELEMETRY TEMPO CORTEX LOKI KEYCLOAK OPA
DOCUMENT REVIEW / PDF

Architecture outcomes described

04 AREAS
01
TELEMETRY COVERAGE

Three observability signals

The PDF describes OpenTelemetry for collection, Tempo for traces, Prometheus/Cortex for metrics, and Fluent Bit/Loki for logs.

IN PDF
02
DASHBOARDS & INVESTIGATION

Cross-signal troubleshooting

Page 9 describes Grafana dashboards combining telemetry and an investigation path from a latency alert to a trace and related logs.

IN PDF
03
ALERTING

Threshold-based alert design

Prometheus Alertmanager is described as issuing metrics-based alerts. Actual rule files, configured thresholds and successful firing evidence were not supplied.

DESCRIBED
04
TENANT ISOLATION

Identity and policy enforcement model

Page 8 assigns identity management to Keycloak and authorization to OPA. Working policies and cross-tenant denial tests are not available as evidence.

VERIFY
ENGINEERING REVIEW / 01

Lessons & next improvements

L01

Telemetry consistency

Preserve trace context and consistent tenant metadata across ingestion, storage and investigation.

L02

Isolation verification

Test whether Tenant A is denied access to Tenant B’s traces, metrics, logs and dashboard queries.

L03

Alert validation

Supply real rule configuration and test threshold breaches, notifications and the investigation workflow.

L04

Recovery evidence

Validate backups, failover behavior and recovery procedures rather than relying on the multi-cloud diagram.

PROJECT RESOURCE EXPLORER PDF AVAILABLE
MOMNA SHAHID / CASE STUDY 01

Multi-Tenant SaaS Observability Architecture

The available project source is a 16-page presentation describing telemetry collection, observability backends, tenant-level access control and a proposed AWS/IBM Cloud environment.

Explore project on LinkedIn
DOCUMENTATION MAP 16-PAGE PDF
▾ observability-platform.pdf
├─ ▤ Architecture overview P04
├─ ▤ Distributed tracing P05
├─ ▤ Metrics with Cortex P06
├─ ▤ Tenant-aware logging P07
├─ ▤ Tenant controls and RBAC P08
├─ ▤ Dashboards and alert correlation P09
└─ ▤ Cloud deployment concept P10

Page references identify topics in the supplied presentation. They are not repository paths or proof of a deployed implementation.

TECHNOLOGY INDEX
OpenTelemetry Grafana Tempo Prometheus Cortex Fluent Bit Loki Keycloak OPA Terraform AWS IBM Cloud
!
IMPLEMENTATION EVIDENCE NEEDED

Source configuration, dashboard exports, alert rules, authorization policies and cross-tenant test results have not been supplied with the presentation.

PROJECT HANDOFF

Architecture documented. Evidence is the next step.

The presentation brings together an open-source observability stack, tenant-level access-control concepts and a multi-cloud infrastructure direction. The next step is to connect those design claims to reviewable implementation artifacts and repeatable tests.

01

Inspect the telemetry pipeline

Verify how traces, metrics and logs are collected, labelled, routed and queried. Confirm the actual OpenTelemetry, Tempo, Cortex and Loki configurations.

02

Validate tenant isolation

Review Keycloak identity configuration and OPA policies. Test that a tenant cannot query another tenant’s telemetry through APIs, dashboards or direct backend access.

03

Capture operational evidence

Provide real Grafana dashboard screenshots, alert-rule configuration, a successful alert test and trace-to-log correlation evidence.

04

Verify cloud recovery claims

Review Terraform configuration and cloud resources. Document actual backup and failover tests before describing high availability as demonstrated.

POTENTIAL NEXT ITERATIONS
+

Add repeatable negative authorization tests for every telemetry access path.

+

Record alert firing, notification delivery and trace-to-log investigation with real artifacts.

+

Document tenant-specific SLOs, resource limits, retention settings and recovery objectives.

These are recommended validation steps, not claims that the listed controls or tests have already been completed.

i
PROJECT STATUS

Source available for review: the supplied architecture PDF. A verified source-code repository and independent deployment evidence have not been provided. Deployment status and effective tenant isolation therefore remain unverified from the available material.

Scroll to Top