INDEPENDENT THINKING / ENGINEERING PRACTICE
PORTFOLIO / 202612°58′ N — 77°35′ EPRODUCT · CLOUD · APPLIED AI

ENGINEER BY PRACTICE. BUILDER BY INSTINCT.

Karthick
Kalimuthu.

I turn complex workflows into useful, reliable software.

Enterprise product engineering, cloud systems, and AI-assisted workflows — designed around the real problem, not the latest buzzword.

Explore the work Connect on LinkedIn
BASEBengaluru, India
EXPERIENCE8 years in software engineering
FOCUSProduct → Platform → AI
FIG. 001 / THE ENGINEERPRODUCT · CLOUD · AI
INDEPENDENT THINKING / ENGINEERING PRACTICE
Karthick Kalimuthu
KK / 019KK.ENGINEER
BY PRACTICE
01PRODUCT
ENGINEERING↗
02APPLIED AI
& CLOUD SYSTEMS✳
THOUGHTFUL SYSTEMSBENGALURU, INDIA
01NEXT.JS02TYPESCRIPT03NODE.JS04AZURE05AWS06LANGGRAPH07RAG SYSTEMS08POSTGRESQL09REACT10VERTEX AI11NESTJS12MYSQL13MONGODB14DYNAMODB
01NEXT.JS02TYPESCRIPT03NODE.JS04AZURE05AWS06LANGGRAPH07RAG SYSTEMS08POSTGRESQL09REACT10VERTEX AI11NESTJS12MYSQL13MONGODB14DYNAMODB

00 / INTERACTIVE INDEX

Skip the scroll.
Query the work.

Explore project IDs, stacks, and tools directly. This console runs a small allowlisted command parser — no shell access.

kk://portfolio — local session

visitor@kk:~$ system

KK / PORTFOLIO CONSOLE — ONLINE

Type `help` to explore project IDs and engineering areas.

01 / PROJECT INDEX — 16 ENTRIES

Systems built for
real-world complexity.

From academic platforms and enterprise operations to agentic AI and retrieval systems. Search the catalogue and open a project for its challenge, approach, stack, and tools.

KK / CASE 01EDUCATION / UI STUDY
LOBBY
ACADEMIC OVERVIEW

Learning in
motion.

1,248STUDENTS36CLASSES
✳
01 / 16SYSTEM STUDY / ILLUSTRATION
Education · Multi-stakeholder platformEnterprise

Academic Information Management System

A connected academic platform for schools, students, parents, and institutions to monitor progress, coordinate communication, and improve academic outcomes.

Project notes

THE CHALLENGE

Academic information is often fragmented between schools, families, students, and governing bodies. Syllabus updates, examination schedules, performance tracking, and parent communication need a shared source of truth.

APPROACH

  • Connect parent, student, school, and institution journeys through role-aware workflows.
  • Support academic progress monitoring and communication across affiliated institutions.
  • Organize shared academic information such as syllabi and examination updates.

TECH STACK

ReactNode.jsREST APIsMySQL / SQLRole-based access

TOOLS USED

GitJiraVS CodeAPI testing
KK / CASE 02COMMERCE / UI STUDY
DISTRIBUTION / 08Order overviewIN TRANSIT
PO-2048124 units•••
PO-205186 units•••
PO-205742 units•••
Stock availability78%
✳
02 / 16SYSTEM STUDY / ILLUSTRATION
CRM · Distribution operationsEnterprise

Distributor CRM & Order Management

A distributor-facing CRM for product catalogues, order placement, inventory visibility, sales tracking, and faster coordination between a business and its distribution network.

Project notes

THE CHALLENGE

Manual order coordination and delayed stock updates create avoidable friction between distributors and central operations.

APPROACH

  • Enable catalogue browsing, order submission, and order-status visibility.
  • Provide stock and sales views to support day-to-day distributor decisions.
  • Centralize operational updates to improve communication and order accuracy.

TECH STACK

ReactNode.js / NestJSREST APIsSQLRBAC

TOOLS USED

GitJiraPostmanCI/CD
KK / CASE 03RESEARCH / FORM STUDY
FIELD RESEARCH / FORM 014Survey response
How satisfied are you with the service?
01  Very satisfied ✓02  Satisfied03  Needs improvement
COMPLETION04 / 06
✳
03 / 16SYSTEM STUDY / ILLUSTRATION
CRM · Field and survey workflowsEnterprise

Survey CRM

A CRM workflow for organizing survey records, lead details, ownership, progress states, and follow-up activities.

Project notes

THE CHALLENGE

Survey and lead operations become difficult to manage when records, assignments, and next actions are spread across disconnected tools.

APPROACH

  • Organize survey and lead records into searchable, trackable workflows.
  • Represent ownership, statuses, and follow-ups explicitly.
  • Use structured APIs and validation to support consistent data capture.

TECH STACK

ReactNode.jsTypeScriptSQL ServerREST APIs

TOOLS USED

GitJiraPostmanVS Code
KK / CASE 04ERP / ANALYTICS STUDY
SALES INTELLIGENCEQuarterly performance
₹42.8LREVENUE1,284ORDERS
JANMARMAYJULSEP
✳
04 / 16SYSTEM STUDY / ILLUSTRATION
ERP · Sales operationsEnterprise

Enterprise Sales & Distribution ERP

An integrated business workflow covering sales operations, distribution processes, records, and operational visibility.

Project notes

THE CHALLENGE

Sales and distribution teams need consistent data and clear handoffs across processes that otherwise live in separate systems.

APPROACH

  • Connect related operational records through clear domain boundaries.
  • Support traceable state changes and role-aware actions.
  • Design API and database contracts for maintainable integrations.

TECH STACK

ReactNode.js / NestJSSQL ServerPostgreSQLREST APIs

TOOLS USED

GitJiraPostmanAzure / AWS tooling
KK / CASE 05OPERATIONS / FLOW STUDY
WORK QUEUE / 07Approval sequence
✓Request submittedCOMPLETE
✓Manager reviewCOMPLETE
03Finance approvalPENDING
04Ready to executeQUEUED
✳
05 / 16SYSTEM STUDY / ILLUSTRATION
Enterprise · Workflow automationEnterprise

Enterprise Workflow System

A configurable workflow application for moving business records through approvals, assignments, status changes, and auditable handoffs.

Project notes

THE CHALLENGE

Operational work slows down when ownership, approval paths, and exception handling are unclear.

APPROACH

  • Represent workflow stages and transition rules explicitly.
  • Build role-aware queues and actionable work views.
  • Keep validation, audit history, and exception states visible.

TECH STACK

ReactNestJSTypeScriptPostgreSQL / SQL ServerRBAC

TOOLS USED

GitJiraCI/CDApplication Insights
KK / CASE 06LEARNING / UI STUDY
LEARNING PATH / 04Designing
for scale.
↗
YOUR PROGRESSCourse modules 68%
✳
06 / 16SYSTEM STUDY / ILLUSTRATION
Learning · Enterprise platformEnterprise

Enterprise Learning Management System

A learning platform for organizing learning content, learner progress, course activities, and administrative workflows.

Project notes

THE CHALLENGE

Learning content, progress tracking, and administration need to work together across learner and instructor journeys.

APPROACH

  • Structure course and learning journeys into maintainable modules.
  • Support progress tracking and role-specific experiences.
  • Keep reporting and access controls aligned with platform responsibilities.

TECH STACK

ReactNode.jsTypeScriptSQLREST APIs

TOOLS USED

GitJiraPostmanVS Code
KK / CASE 07RURAL / NETWORK STUDY
HUB 01HUB 02HUB 03STORERURAL
LAST-MILE NETWORK05 DISTRIBUTION POINTS
✳
07 / 16SYSTEM STUDY / ILLUSTRATION
Commerce · Distribution networkEnterprise

Rural Assisted Commerce & Last-Mile Distribution

A last-mile distribution platform concept focused on connecting rural customers, assisted-commerce operators, and fulfilment workflows.

Project notes

THE CHALLENGE

Last-mile commerce requires coordination between assisted access points, product availability, orders, and distribution operations across dispersed communities.

APPROACH

  • Model order and fulfilment journeys for assisted-commerce workflows.
  • Make availability and status changes easier to track.
  • Support operational visibility across the distribution chain.

TECH STACK

ReactNode.jsREST APIsSQL / NoSQLCloud services

TOOLS USED

GitJiraAPI testingCloud monitoring
KK / CASE 08MEDICAL / SIMULATION STUDY
SIMULATION LABSESSION 03 / ACTIVE
SIM
PROCEDURE STEPS01  Prepare02  Observe03  Practice
TRAINING ENVIRONMENT●  SAFE MODE
✳
08 / 16SYSTEM STUDY / ILLUSTRATION
Freelance · Medical educationFreelance

Medical E-Learning & Practical Simulation

A university e-learning experience with practical, simulation-based learning activities for medical students, including virtual animal-anatomy and procedure practice.

Project notes

THE CHALLENGE

Medical learners need structured theory and practical rehearsal opportunities in a controlled digital learning environment.

APPROACH

  • Connect learning material with step-by-step practical activities.
  • Present procedure stages and learning feedback in a clear sequence.
  • Design the experience around educational use rather than claiming clinical validation.

TECH STACK

Web frontendJavaScript / TypeScriptBackend APIsSQL

TOOLS USED

GitVS CodeBrowser debuggingAPI testing
KK / CASE 09HEALTHCARE / UI STUDY
H+
PATIENT RECORD / DEMO
P
Patient summaryRecord # 00824
ACTIVE
98.6°TEMP.72HEART RATE120/80BP
Clinical notesUPDATED
Medication historyREVIEWED
✳
09 / 16SYSTEM STUDY / ILLUSTRATION
Healthcare · Application workflowsEnterprise

Electronic Health Record System

Healthcare application workflows for organizing patient-related records and operational information with attention to privacy, traceability, and data integrity.

Project notes

THE CHALLENGE

Healthcare systems need reliable data handling, controlled access, and predictable workflows across sensitive records.

APPROACH

  • Separate domain APIs and data access boundaries.
  • Design role-based permissions and audit-aware operations.
  • Treat validation, migration safety, and privacy as core requirements.

TECH STACK

React / Next.jsNode.js / NestJSSQL Server / PostgreSQLREST APIs

TOOLS USED

GitJiraPostmanTesting and monitoring
KK / CASE 10PROCUREMENT / DOCUMENT STUDY
PURCHASE
REQUESTPR-1048
REQUESTOROperations teamESTIMATED COST₹ 1,28,000
MANAGER REVIEW✓ APPROVED
APPROVAL
TRACKING
✳
10 / 16SYSTEM STUDY / ILLUSTRATION
Enterprise · ProcurementEnterprise

Contractor Management & Purchase Requests

An end-to-end contractor management workflow with purchase-request initiation, review, approvals, and operational tracking.

Project notes

THE CHALLENGE

Contractor and procurement processes cross teams, approvals, and supporting records, making ownership and status visibility important.

APPROACH

  • Connect contractor records with request and approval journeys.
  • Model approval states and responsible roles explicitly.
  • Provide traceability from request creation through outcome.

TECH STACK

ReactNestJSTypeScriptSQL Server / PostgreSQLRBAC

TOOLS USED

GitJiraPostmanCI/CD
KK / CASE 11DATA / REPORT STUDY
ASSESSMENT DATABASEGlobal response quality
94.2%
VALIDATED RECORDS+8.4% vs previous cycle
REGIONRECORDSSTATUS
APAC2,480VERIFIED
EMEA1,920VERIFIED
AMER3,102REVIEW
✳
11 / 16SYSTEM STUDY / ILLUSTRATION
Data platform · Assessment operationsEnterprise

Global Confirmatory Assessment Database

A database-backed assessment system for maintaining confirmatory assessment records and supporting consistent retrieval and operational review.

Project notes

THE CHALLENGE

Assessment records need consistent structure, reliable querying, and appropriate access across distributed users and workflows.

APPROACH

  • Design relational structures around assessment entities and relationships.
  • Support validated record capture and retrieval.
  • Keep permissions and data consistency central to the application design.

TECH STACK

SQL ServerMySQL / PostgreSQLBackend APIsData validation

TOOLS USED

SQL toolingGitJiraAPI testing
KK / CASE 12AI / PIPELINE STUDY
AGENTIC SDLC / EXECUTION TRACERequirement → release
01ANALYSEDONE
02ARCHITECTDONE
03IMPLEMENTDONE
04TESTREADY
ORCHESTRATION● ALL SYSTEMS CONNECTED
✳
12 / 16SYSTEM STUDY / ILLUSTRATION
Applied AI · Software deliveryAI / R&D

Agentic SDLC Automation

A multi-agent workflow that assists with requirement analysis, architecture artefacts, documentation, issue creation, code scaffolding, design flows, and test-case generation.

Project notes

THE CHALLENGE

Requirement changes can trigger repeated manual cycles across analysis, design, documentation, backlog creation, implementation, and testing.

APPROACH

  • Separate agent responsibilities into inspectable stages.
  • Use orchestration to coordinate outputs and tool-enabled steps.
  • Keep generated artefacts reviewable by engineers before they are treated as final.

TECH STACK

Node.js / NestJSLangGraphLLM APIsGitHub / Jira integrationsTypeScript

TOOLS USED

LangChainLangGraphGitHubJiraPrompt evaluation
KK / CASE 13FINOPS / ANALYTICS STUDY
CLOUD COST INTELLIGENCEMonthly spend
$42,850↓ 12.8% vs last month
COMPUTE STORAGE NETWORK
✳
13 / 16SYSTEM STUDY / ILLUSTRATION
Applied AI · Cloud cost intelligenceAI / R&D

Argus AI — Multi-Agent FinOps

A multi-agent FinOps solution exploring cloud-billing analysis, anomaly investigation, root-cause hypotheses, and reviewable remediation suggestions.

Project notes

THE CHALLENGE

A cost spike is easy to spot but harder to explain without connecting billing records to resources and operational context.

APPROACH

  • Use billing data as the evidence layer for cost investigation.
  • Coordinate agents to investigate potential causes and summarize findings.
  • Keep remediation proposals traceable and subject to human review.

TECH STACK

Google CloudBigQueryVertex AI / GeminiPython / FastAPIAgent orchestration

TOOLS USED

Vertex AIBigQueryLangGraph conceptsGitObservability
KK / CASE 14ELECTRONICS / SCHEMATIC STUDY
DESIGN ASSURANCE / BOARD 07Signal integrity map
NETLISTRULE CHECKREPORT
✳
14 / 16SYSTEM STUDY / ILLUSTRATION
Applied AI · Electronics engineeringAI / R&D

Engentica-Inspired Electronics Design Assurance

A design-assurance platform concept using vision-enabled AI and deterministic analysis to inspect electronic design artefacts and surface reviewable findings.

Project notes

THE CHALLENGE

Engineers need traceable findings that connect visual design context, parsed structure, and rule-based checks without relying on an opaque model response.

APPROACH

  • Combine vision-based interpretation with structured parsing where possible.
  • Run deterministic validation rules before generating AI-assisted explanations.
  • Link findings to source artefacts and clearly communicate confidence and limitations.

TECH STACK

PythonVision-language modelsNetlist / schematic parsingRule engineRAG

TOOLS USED

Local LLM inferenceOllama / inference toolingQdrantGitEvaluation
KK / CASE 15KNOWLEDGE / RAG STUDY
Search knowledge base…⌕
Policy handbookPDF / 12 pages
Technical guideDOCX / 8 pages
Release notesPDF / 4 pages
GENERATED RESPONSE / SOURCEDAnswer with context.[1] [2] SOURCES
RETRIEVE→RERANK→GENERATE
✳
15 / 16SYSTEM STUDY / ILLUSTRATION
Applied AI · Enterprise knowledgeAI / R&D

RAG-Based Knowledge Assistant

A retrieval-augmented chatbot that answers questions using indexed source documents, with a focus on grounded responses and access-aware retrieval.

Project notes

THE CHALLENGE

A chatbot is not useful for enterprise knowledge if it cannot cite source material, respect data boundaries, or acknowledge missing evidence.

APPROACH

  • Parse, chunk, embed, and index source content for retrieval.
  • Retrieve relevant passages and ground answers in available context.
  • Plan access filters, source references, evaluation, and fallback behavior into the pipeline.

TECH STACK

Azure AI Search / QdrantNode.jsLLM orchestrationEmbeddingsRAG

TOOLS USED

LangChainLangGraphQdrantOpen WebUILangfuse
KK / CASE 16AGENTIC AI / SYSTEM STUDY
CONFIGURABLE AGENT SYSTEMChoose a capability.
01PLAN
02SEARCH
03TOOLS
04MEMORY
05EVALUATE
06RESPOND
CONFIGURATION-DRIVENTOOLS / MEMORY / GUARDRAILS
✳
16 / 16SYSTEM STUDY / ILLUSTRATION
Applied AI · Modular platformAI / R&D

Configurable Agentic AI Solutions

A modular AI application architecture for composing agents, RAG, document parsing, tools, prompt management, and provider integrations through clear module contracts.

Project notes

THE CHALLENGE

AI projects become difficult to extend when orchestration, retrieval, parsing, providers, prompts, and user interfaces are tightly coupled.

APPROACH

  • Separate agent orchestration, RAG, document processing, and provider adapters.
  • Use configuration and contracts to keep modules replaceable and testable.
  • Plan security, observability, evaluation, and deployment constraints from the start.

TECH STACK

Next.jsNestJSLangGraphAzure AI SearchAzure SQL / PostgreSQL

TOOLS USED

Microsoft FoundryManaged IdentityLangChainLangfuseN8NQdrantRedisGitHub Actions

PROJECT SUMMARIES ARE HIGH-LEVEL PORTFOLIO DESCRIPTIONS / NO CONFIDENTIAL COMPANY NAMES INCLUDED.

02 / TECHNICAL PERIODIC TABLE

Every tool has
a place.

Skills grouped as a working system — frontend, backend, databases, applied AI, cloud, and developer tooling. Tap a tile to jump into the related project search.

UI

Frontend

React · Next.js · Angular · TypeScript · JavaScript · Redux · Tailwind CSS

BE

Backend

Node.js · NestJS · Express.js · FastAPI · REST APIs · Prisma · TypeORM

DB

Databases

MySQL · PostgreSQL · Microsoft SQL Server · MongoDB · DynamoDB · Azure SQL · Redis

AI

AI / orchestration

RAG · LangChain · LangGraph · Vertex AI · Gemini · Microsoft Foundry · Semantic Kernel · Ollama

CL

Cloud & DevOps

Azure · AWS · Google Cloud · App Service · S3 · Lambda · Docker · GitHub Actions

DX

Developer tools

Git · GitHub · Jira · VS Code · Postman · N8N · Qdrant · Langfuse · Open WebUI · Application Insights

03 / THE WORKBENCH

Tools behind
the shipped work.

Infrastructure and engineering tools I have worked with across application development, cloud, automation, retrieval, and local AI experimentation.

AI

AI & agent orchestration

Agents, graphs, model orchestration

12 TOOLS
LangGraph
LangChain
Semantic Kernel
GenKit
Microsoft Foundry
Vertex AI
Gemini
Ollama
Qwen
Mistral
DeepSeek
Gemma
RAG

Retrieval & knowledge

Search, embeddings, document pipelines

05 TOOLS
Azure AI Search
Document Intelligence
Qdrant
Langfuse
Open WebUI
UI

Frontend engineering

Web applications and component systems

08 TOOLS
Next.js
React
Angular
TypeScript
JavaScript
Tailwind CSS
Redux
Zustand
API

Backend & APIs

Services, contracts, application logic

07 TOOLS
Node.js
NestJS
Express.js
FastAPI
Python
Prisma
TypeORM
DB

Data platforms

Relational, document, and key-value data

07 TOOLS
PostgreSQL
MySQL
Microsoft SQL Server
MongoDB
DynamoDB
Redis
BigQuery
CLD

Cloud & platform

Hosting, storage, monitoring, infrastructure

09 TOOLS
Azure
AWS
Google Cloud
Azure App Service
Azure Functions
Azure Blob Storage
Azure Resource Graph
Azure Monitor
Docker
DEV

Delivery & collaboration

Source control, automation, engineering workflow

09 TOOLS
Git
GitHub
GitHub Actions
VS Code
Jira
Confluence
n8n
Postman
Swagger / OpenAPI
OPS

Security, testing & operations

Identity, access, quality, observability

10 TOOLS
Microsoft Entra ID
Managed Identity
RBAC
Key Vault
Application Insights
Playwright
Jest
Vitest
FinOps
Observability

04 / ENGINEERING APPROACH

Make the hard parts
less mysterious.

Good engineering is not the number of technologies on a page. It is how well the system handles real workflows, changing requirements, security, and failure.

01

Understand the work

Start with users, constraints, operational bottlenecks, and what success should mean.

02

Design clear boundaries

Make data contracts, API responsibilities, permissions, and failure paths explicit.

03

Build for change

Keep modules testable and integrations replaceable. Avoid architecture that is clever but fragile.

04

Prove it works

Use tests, telemetry, reviewable AI outputs, and real evidence rather than unsupported claims.