Shubh Kamal Sharma

Full-Stack SDE & AI Builder

Hi, I'm Shubh Kamal Sharma

Full-stack engineer who takes products from design to deployment — distributed backends, multi-agent AI pipelines, and cloud-native infrastructure built to run reliably in production.

Currently at Microsoft | Previously at Prodevans & ModelX AI

About

Engineering at the intersection of AI & Scale

Product-focused engineer with 2+ years at Microsoft, Prodevans Technologies & ModelX AI, shipping end-to-end systems across distributed backends, AI pipelines, and cloud-native infrastructure.

I started as a developer obsessed with how large systems stay reliable under pressure — that curiosity led me from building Spring Boot REST APIs and React UIs to designing event-driven Kafka pipelines and multi-agent AI workflows with LangChain and CrewAI.

At Prodevans Technologies I architected data pipelines using Kafka and InfluxDB, improved system latency with Redis caching strategies, and automated CI/CD with Jenkins and Kubernetes. At Microsoft I build and integrate Azure-native microservices, work on API gateway and Functions-based services, and drive observability efforts across distributed systems.

Beyond work I ship products I care about — FinSight AI (a CrewAI multi-agent financial document analyser), AgenticFlow (a visual DAG orchestration platform), and platforms like CloudOps Hub and JobSight AI. I believe in full-stack ownership: from the first API design to the post-mortem.

AI & Agents

LangChain, LangGraph, RAG pipelines, multi-agent orchestration with HITL checkpoints.

Full Stack

React, Next.js, FastAPI, NestJS, Spring Boot — end-to-end product delivery.

Cloud Native

AWS, Azure, Docker, Kubernetes, Terraform — production-grade infra at scale.

Performance

Kafka event-driven pipelines, Redis caching, and async job queues — built for throughput and low latency.

Projects

What I've been building

Production-grade systems at the intersection of full-stack engineering and agentic AI.

AI-powered products — multi-agent pipelines, agentic orchestration, LLM integrations.

Financial Analysis

Source

FinSight AI

Multi-Agent Financial Document Analyzer

AGENTIC AICREWAIREST API

Full-stack app to upload financial PDFs and generate structured outputs — executive summaries, KPIs, and risk flags — via a secure FastAPI REST API with JWT + RBAC. A CrewAI multi-agent pipeline (financial analyst, risk assessor, and report generator agents) handles analysis, while Celery workers backed by Redis handle async processing.

  • Multi-agent analysis pipeline built with CrewAI (gpt-4o-mini via LangChain) — financial analyst, risk assessor, and report generator agents collaborate per document.
  • Async job queue: 6 Celery task types process uploads, analysis, and exports independently with live progress tracking via status endpoint.
  • Two-tier Redis cache (hot-path summaries + session data) reduces redundant LLM calls on repeat documents.
  • Secured with JWT authentication and role-based access control (RBAC); supports PDF uploads up to 50 MB.
PythonFastAPIReactTypeScriptMongoDBRedisCeleryCrewAILangChainOpenAI

AI Orchestration

Source

AgenticFlow

Visual Multi-Agent Workflow Orchestrator

AGENTIC AIKAFKAREAL-TIME

Production-ready orchestration platform with a visual drag-and-drop workflow builder (React Flow) for composing deterministic DAGs alongside LLM-powered agentic nodes. Event-driven backend with Kafka (workflow.events, node.events, agent.events topics), PostgreSQL state snapshots, and real-time WebSocket monitoring.

  • Event-driven backend powered by Kafka (workflow.events, node.events, agent.events) — async node execution with Redis distributed locking and retry logic.
  • Visual drag-and-drop DAG builder using React Flow for composing agentic workflows with deterministic and LLM-powered nodes.
  • Multi-turn Human-in-the-Loop (HITL) checkpoints with state snapshots and workflow replay built in.
  • Supports GPT-4 and Claude 3 as interchangeable LLM backends via LangChain.
  • Kubernetes-orchestrated deployment with real-time WebSocket streaming for live execution monitoring.
PythonFastAPINext.jsReact FlowTypeScriptPostgreSQLRedisKafkaKubernetes

AI SaaS Platform

Enterprise AI Agent Builder

Visual Platform to Compose, Deploy & Monitor Production AI Agents

REACT FLOWMULTI-TENANTREAL-TIME

UI engineering on an enterprise SaaS platform for composing, deploying, and monitoring production AI agents across multi-tenant organisations. Owned delivery of the visual agent flow builder (React Flow canvas), in-browser agent playground with real-time SSE streaming, knowledge base management UI, and a live observability dashboard — each wired to REST APIs with Zustand-managed client state.

  • Visual agent flow builder on a React Flow canvas — engineered drag-and-drop composition of LLM nodes, tool-call components, and conditional branches with custom node renderers and canvas state synced bidirectionally with the backend API.
  • In-browser agent playground with real-time SSE streaming — prompt iteration, live step-by-step execution inspection, and error surfacing before agents are promoted to production.
  • Knowledge base management UI — document upload, listing, search, and per-agent attachment flows with optimistic UI updates and full API integration.
  • Live observability dashboard — usage metrics, latency charts, and execution history with high-frequency state updates and layout-stable rendering.
  • Multi-tenant org and RBAC management — team, department, role assignment, prompt gallery, and leaderboard; reusable MUI component system with Zustand across all admin flows.
Next.jsTypeScriptReact FlowMUIMonaco EditorZustandnext-authnext-intlDocker

AI Recruiting

AI Job Search & Matching Platform

Multi-Agent Engine Aggregating 6+ Job Sources with Smart Ranking

MULTI-AGENTAWS CDKMOBILE

AI job search platform with a 4-agent pipeline (Coordinator → QueryParser, Fetcher, CurationAgent) that parses natural language queries, aggregates listings from 6+ job APIs, deduplicates via similarity clustering, and ranks results using a 7-factor weighted scoring model. Deployed on AWS via CDK-managed infrastructure and packaged as a native Android app via Capacitor.

  • 4-agent pipeline: CoordinatorAgent orchestrates QueryParserAgent (query expansion + synonyms), FetcherAgent (multi-source parallel fetching), and CurationAgent (deduplication + ranking).
  • Aggregates listings from 6+ sources — Adzuna, JSearch, CoreSignal, Jooble, RemoteOK, Arbeitnow — with similarity-based deduplication across all results.
  • 7-factor relevance scoring: title match (25%), skills match (30%), experience, location, company reputation, job freshness, and description relevance.
  • Resume upload with PDF parsing for personalised job matching based on extracted skills and experience.
  • Full AWS CDK deployment: VPC, RDS (PostgreSQL), S3, Cognito auth, ALB, and ECS Fargate — environment-specific stacks for dev and prod.
PythonFastAPINext.jsTypeScriptOpenAIAWS CDKECS FargateCognitoRDS / Aurora

Cloud Governance

AI Cloud Infrastructure Platform

Multi-Cloud Catalog & IaC Console with Conversational AI

AI AGENTMULTI-CLOUDENTERPRISE

End-to-end UI engineering on an enterprise cloud infrastructure management platform — delivered the multi-cloud service catalog, versioned IaC blueprint viewer with an embedded code editor, AI chat panel with streamed responses, and data-heavy org and RBAC management flows, making complex infrastructure tooling accessible to non-DevOps teams.

  • Multi-cloud service catalog — engineered categorised AWS, Azure, and GCP service views with tabbed detail layouts (overview, features, security, resources), locale-aware routing, and API-driven content loading.
  • Versioned IaC blueprint viewer — interactive code editor integration (Monaco) for browsing Terraform, YAML, Python, Go, and JSON blueprints with version switching and diff-ready layout.
  • AI chat panel with streamed response rendering — conversational UI for natural language infrastructure queries, integrated with a backend agent service via streaming API calls and incremental UI updates.
  • Multi-tenant org and RBAC management — business units, departments, teams, cloud account configs, and policy assignment flows; complex form handling, optimistic updates, and role-gated route access.
  • Multi-locale internationalisation (next-intl) across all protected and public routes with RTL layout support and locale-aware navigation — shipped as a cross-cutting concern across the entire product.
Next.jsTypeScriptReactTailwindCSSPostgreSQLDocker

Open Source & More

More on GitHub

Microservices, AI tools, and side projects across the full stack.

13+

GitHub repositories

4

AI agent systems

3

Cloud platforms built

View GitHub
Experience

Professional Journey

Building at the intersection of full-stack engineering and agentic AI — focused on measurable impact, observability, and end-to-end ownership.

Oct 2025 — Jun 2026

Software Engineer

Microsoft· Contract

Bengaluru, India

  • Develop and integrate backend microservices using Azure API Gateway, Azure Functions, and enterprise REST APIs within an agile engineering team.
  • Build and ship automation scripts and CI/CD pipelines (GitHub Actions, Docker) following production best practices: structured logging, zero-downtime deployment.
  • Drive observability efforts by analysing distributed telemetry and logs, authoring runbooks, and contributing to incident post-mortems — reducing MTTR on recurring issues.
  • Collaborate cross-functionally on system integration projects; deliver reviewed PRs and technical documentation that accelerated onboarding for new engineers.
Azure API GatewayAzure FunctionsGitHub ActionsDockerREST APIsObservability

Dec 2024 — Aug 2025

Associate Software Engineer

Prodevans Technologies Pvt Ltd· Full-time

Bhubaneswar, India

  • Engineered full-stack features across React/Next.js (TypeScript) frontends and Spring Boot/NestJS backends; designed REST and GraphQL APIs for multi-tenant workflows with RBAC.
  • Architected stateless, event-driven data pipelines using Kafka and InfluxDB, processing 100k+ events/day; reduced end-to-end latency by ~30% via Redis caching and intelligent batching.
  • Automated CI/CD workflows with Jenkins and GitHub Actions; containerised services with Docker and orchestrated deployments via Kubernetes — cutting release time by ~40%.
  • Delivered end-to-end user-facing features with a focus on UX quality and system observability; integrated monitoring dashboards tracking key service health metrics across production.
ReactNext.jsTypeScriptSpring BootNestJSGraphQLKafkaInfluxDBRedisKubernetesJenkins

Apr 2024 — Nov 2024

Freelance Software Engineer

ModelX AI· Freelance

Remote (US-based startup)

  • Engineered core UI features across a multi-tenant enterprise AI agent platform — delivered the visual agent flow builder (React Flow canvas with drag-and-drop LLM nodes, tool calls, and conditional branches), real-time observability dashboard, and knowledge base management, integrating each with REST APIs and Zustand-managed client state.
  • Owned end-to-end UI delivery for a cloud infrastructure management product — implemented the multi-cloud service catalog (AWS, Azure, GCP), versioned blueprint viewer with an embedded code editor, AI chat panel with streamed responses, and data-heavy org and RBAC management flows.
  • Architected multi-locale internationalisation (next-intl) across all routes — locale-aware routing, RTL layout support, and dynamic namespace loading — directly expanding the platform's addressable market to non-English enterprise customers.
  • Collaborated asynchronously with a US-based founding team across sprint cycles; drove frontend architecture decisions, maintained component standards, and shipped production-quality PRs with thorough code review.
Next.jsTypeScriptReact FlowMUIMonaco EditorZustandnext-authnext-intlDocker
Skills

Technical toolkit

Tools and technologies I've shipped in production across full-stack, GenAI, and cloud-native systems.

GENAI & AGENTS

LangChainLangGraphCrewAILangSmithOpenAI (GPT-4)Anthropic ClaudeAWS BedrockRAG SystemsVector DatabasesEmbeddingsPrompt EngineeringAgentic WorkflowsFine-Tuning

LANGUAGES

TypeScriptJavaScriptPythonJavaGoC#Node.jsSQL

FRONTEND

ReactNext.jsAngularReact NativeTailwindCSSReact FlowRedux ToolkitMUIZustand

BACKEND & APIS

FastAPIExpress.jsNestJSSpring Boot.NETRESTGraphQLWebSocketJWT / OAuth2MicroservicesSystem DesignEvent-Driven Architecture

CLOUD & INFRA

AWS LambdaECS FargateDynamoDBRDS / AuroraS3 + CloudFrontAPI GatewayCognitoAzureDockerKubernetesTerraformAWS CDKGitHub ActionsJenkins

DATA & MESSAGING

PostgreSQLMongoDBRedisKafkaSQSOpenSearchCeleryInfluxDB

Education

Bachelor of Technology in Computer Science and Engineering

Dr. A. P. J. Abdul Kalam Technical University

Lucknow, India · August 2024

Certifications

  • LangChain — Develop AI Agents with LangChain & LangGraph

    Udemy

  • Foundations of User Experience (UX) Design

    Google

  • Java Programming Certification

    Udemy

  • Cloud Engineer Professional Certificate

    Coursera

Contact

Let's connect

Have a project in mind, want to collaborate, or just want to say hi? I'd love to hear from you.

Email

shubhpersonalok@gmail.com

Phone

+91-8303031741

Location

Bengaluru, India

LinkedIn

linkedin.com/in/shubh23

Send a Message

Currently open to new opportunities