ENTERPRISE INNOVATION LAB

PalyamIQ

Enterprise Integration. ERP Intelligence. Observability. Applied AI.

An innovation lab exploring enterprise integrations, ERP intelligence, observability, governance, agentic systems, and applied AI through architecture, experiments and proof-of-concepts.

ERP Intelligence iPaaS & Integrations Enterprise Architecture Observability Applied AI
Available for freelance engagements — OIC & Fusion Middleware · ERP & SCM · SoD/RBAC · AI-Assisted Automation
Explore Projects
SoD Governance · Threat Intelligence · ERP Security · Pipelines · CI/CD · Workflows · API Orchestration · Microservices · AI/LLM Automation · Event-Driven Architecture · iPaaS SoD Governance · Threat Intelligence · ERP Security · Pipelines · CI/CD · Workflows · API Orchestration · Microservices · AI/LLM Automation · Event-Driven Architecture · iPaaS

Why This Lab

Enterprise systems were
never designed to think
together.

I kept seeing the same pattern across enterprise systems: integrations moved data, monitoring tools generated alerts, and business applications captured critical events — but the context rarely traveled with them.

The challenge wasn't collecting information. It was connecting it.

PalyamIQ is an innovation lab exploring enterprise integrations, ERP security, observability, and applied AI through hands-on projects and experiments.

The goal is simple: explore how connected systems can deliver better visibility, stronger governance, and more intelligent automation.

Core Expertise

Enterprise Architecture & Integration

Oracle Fusion ERP

ERP integrations, reporting, governance and business process automation.

Oracle Integration Cloud

Enterprise integration architecture, APIs, event-driven systems and observability.

Applied AI

AI governance, AgentOps, ERP intelligence and enterprise automation.

9
Innovation Modules
Fusion Event Intelligence, Threat Intelligence, iPaaS GenAI Governance, ERP Security, NASA EONET, Applied AI, AgentOps, Odoo AI ERP Intelligence, AI Authentication & Authorization
1
Unified Architecture
Connected pipelines across governance, security, and automation domains.
Open
Research & Prototypes
Research projects, prototypes, and technical explorations.
∞
Continuous Research
Ongoing exploration. Continuous iteration. Always learning.

The System

Nine modules.
One intelligence
layer.

CS-09
FastAPI JWT/OAuth2 Multi-Tenant

AI Authentication & Authorization

A production-pattern security layer for multi-tenant AI/LLM APIs — hashed API keys, JWT-based OAuth 2.0, tenant-tiered access control, and compliance audit logging, adversarially tested end-to-end.

Problem

Shared AI infrastructure needs machine-identity auth, tenant isolation, and cost-trackable audit trails before it can serve more than one customer safely.

Approach

Built and ran 17 success/failure tests on live AWS infrastructure across four modules, including deliberate adversarial attacks against each one.

Outcome

16 of 17 tests passed as designed; testing surfaced a real, reproducible refresh-token reuse gap — found, confirmed three times, and documented rather than hidden.

Published July 2026

CS-08
Odoo 19 AI Copilot

Odoo AI ERP Intelligence

AI-native Enterprise Intelligence Layer on Odoo 19 CE. Real-time workflow monitoring, predictive SLA breach detection, and AI copilot.

Problem

ERP operators lacked predictive visibility into SLA breaches and workflow bottlenecks.

Approach

Built predictive models on Odoo transaction data with AI copilot for natural language queries.

Outcome

92% accurate SLA breach prediction turns reactive firefighting into early intervention, with conversational access for non-technical operators.

Published May 2026

CS-07
LangGraph Langfuse

LLM Observability Platform

Enterprise-grade AI cost control, agent observability, and human-in-the-loop governance for multi-team LLM deployments.

Problem

Multi-agent workflows consumed 3x budget without tracking. No cost attribution per team.

Approach

Built per-team cost gates, approval flows, and Langfuse tracing across agent graphs.

Outcome

Real-time spend visibility caught runaway AI costs before they hit budget, with full audit trails for governance sign-off.

Published May 2026

CS-06
BERT NER

Teaching a Model to Read Technical Changelogs

Building a model that understands dense technical changelogs: what changed, what broke, and why off-the-shelf AI was not enough.

Problem

Generic NER models missed domain-specific terms and severity classifications entirely.

Approach

Fine-tuned a domain-adapted BERT model on technical changelog text with custom NER labels.

Outcome

Extraction accuracy jumped from 34% to 89%, making automated changelog triage viable instead of a manual read-through.

Published January 2026

CS-05
Data Pipelines Geospatial NASA APIs

Environmental Intelligence

Environmental intelligence using NASA EONET data pipelines for natural event monitoring and geospatial risk analysis.

Problem

Disaster event data was fragmented across sources with no unified geospatial view.

Approach

Built unified EONET pipeline with ISRO augmentation and real-time geospatial rendering.

Outcome

Single-pane global monitoring with correlated risk scoring replaced manual cross-referencing across disconnected sources.

Published February 2026

CS-04
Risk Scoring SoD

SoD Governance

Rule-based SoD generates too much noise. A smarter scoring approach caught control failures our old spreadsheet missed.

Problem

Static rule-based SoD produced hundreds of false positives. Real risks were buried.

Approach

Replaced binary rules with weighted risk scoring and peer deviation analysis.

Outcome

Noise cut 80%, so audit teams spend their time on real control failures instead of chasing false positives.

Published March 2026

CS-03
AI Governance iPaaS

Auditing AI-Built Integration Flows

LLMs generate iPaaS mappings fast. They also hallucinate endpoint names. Here is how we guard every AI-generated flow before production.

Problem

Generated flow hit a production PII endpoint during testing. No validation gate existed.

Approach

Added endpoint whitelist validation and schema enforcement before any deployment.

Outcome

100% of AI-generated flows now clear automated governance checks before production, closing the PII exposure risk entirely.

Published May 2026

CS-02
Cybersecurity MITRE ATT&CK Geospatial

Cyber Threat Intelligence

Real-time cyber threat monitoring with MITRE ATT&CK mapping, IOC correlation, and geospatial threat origin analysis.

Problem

Initial threat feed ingestion hit API rate limits. Dashboard updates were intermittent.

Approach

Added caching layer and batch processing with backoff strategy for threat feeds.

Outcome

Reliable 60-second refresh with zero rate-limit violations turned an unstable feed into a dashboard analysts can actually trust.

Published April 2026

CS-01
iPaaS Kafka

Streaming ERP Events Without Losing Your Mind

How iPaaS → Kafka cut event latency, why the DLQ saved us at 2 AM, and the REST introspection trick no one documents.

Problem

Synchronous REST calls collapsed at 200 events/sec. Lost 4 hours of order events.

Approach

Switched to async event streaming with dead-letter queue and REST introspection.

Outcome

Zero event loss and sub-second latency eliminated a recurring source of missed orders, with full observability into ERP event flows.

Published May 2026

The Architecture

Connected Intelligence.
Unified Systems.

01

Integration Intelligence

Connect enterprise applications, APIs, events, and workflows across distributed systems.

02

Access Governance

Identify access risks and policy violations across enterprise systems.

03

Event Intelligence Layer

Transform enterprise events into real-time insights using event-driven architectures.

04

AI-Powered Automation

Apply AI and intelligent workflows to monitoring, decision support, and operations.

Unified Data Flow

Enterprise Systems → Integration Layer
ERP Governance → Risk Intelligence
Event Streams → Operational Insights
AI/LLM Layer → Automation & Decisions

Capabilities

Technical
Focus Areas.

Adaptive Anomaly Detection

Context-aware anomaly detection for enterprise systems, moving beyond static threshold-based checks.

Threat-Aware Governance

Connecting external threat intelligence with internal access policies for faster incident response.

Continuous Compliance Monitoring

Real-time visibility across governance scans, threat feeds, and operational events.

Supply Chain Intelligence

Demand forecasting and logistics optimization for inventory and distribution networks.

Technical Insights

Notes from the Lab.

Integration Patterns

Integration Error Handling Patterns

Structured fault handling, retry strategies, dead-letter queues, and resilience patterns for modern integration platforms.

ERP Strategy

ERP Reporting & Analytics Patterns

Choosing the right reporting approach for operational, analytical, and real-time ERP data needs.

Architecture

Low-Latency Analytics Layers

Notes on building lightweight analytics layers that observe transactional systems without slowing them down.

AgentOps

AgentOps for Enterprise AI

Cost gates, tracing, governance, and human-in-the-loop patterns for production AI agent deployments.

Observability

Observability for Integration Platforms

Unified telemetry across iPaaS, middleware, APIs, and SaaS endpoints using OpenTelemetry and Prometheus.

Identity

About PalyamIQ

I'm Kasu Mallikarjuna, an Integration Architect exploring how enterprise systems connect, operate, and evolve through integration, observability, governance, and applied AI.

PalyamIQ is my innovation lab and technical portfolio, documenting projects, experiments, and proof-of-concepts across enterprise architecture, ERP intelligence, AI systems, and operational visibility.

Integration Systems

OIC · Dell Boomi · SOA Suite · OSB · REST/SOAP · Kafka

ERP Security

SoD Governance · RBAC · Role Design · Access Controls · Audit Analytics

Applied Intelligence

Risk Intelligence · Connected Insights · Context-Aware Automation · Decision Systems