> system online — hello, I'm

Aman Kumar.

I build |

AI Software Engineer with 3+ years of experience shipping production-grade AI platforms, multi-agent systems, enterprise RAG, and real-time Voice AI — for Fortune 500 enterprises and 0-to-1 startups alike.

View my work Get in touch

LangGraph · MCP

RAG · Voice AI

0 Years experience

0 AI inference cost cut

0 Platform uptime

0 Production APIs shipped

01. About Me

I'm an AI Software Engineer who loves turning ambitious AI ideas into systems that actually run in production. Over the last 3+ years I've gone from being the sole founding backend engineer of an AI SaaS startup to building enterprise AI platforms for a Fortune 500 company.

My sweet spot is the intersection of agentic AI and solid backend engineering — designing multi-agent systems with LangChain, LangGraph, and the Model Context Protocol (MCP), building enterprise RAG pipelines that ingest anything from PDFs to spreadsheets, and wiring up real-time Voice AI with LiveKit, WebRTC, and ElevenLabs. Underneath it all: Python (FastAPI), Java (Spring Boot), and cloud infrastructure across AWS, Azure, and GCP.

I care about the unglamorous parts that make AI products dependable — observability with LangSmith and Datadog, latency SLOs, cost-aware model routing, and infrastructure as code with Terraform.

A few things I've done recently:

Cut AI inference costs 45% with intelligent LLM routing
Delivered 99.99% uptime as a solo founding backend engineer
Built AI agents for VS Code, Cursor, Claude Code & IntelliJ
Led a cross-functional AI + backend + frontend team

02. Where I've Worked

Building production-scale AI platforms for a Fortune 500 enterprise with Python (FastAPI) and microservices.
Developed custom AI agents for VS Code, Cursor, Claude Code, and IntelliJ IDEA — automating code generation, repo analysis, and SDLC workflows, improving developer productivity by 45%.
Designed multi-agent systems with LangChain, LangGraph, and MCP, integrated with Azure DevOps and GitHub.
Engineered intelligent LLM routing (cheap models for simple tasks, premium for hard reasoning), cutting inference costs by 45%.
End-to-end AI observability with LangSmith & Datadog: prompt versioning, tracing, evals, P95 latency SLOs.

FastAPILangGraphMCPRAGTerraformAzure

Built a production-scale AI Agent Platform where users create, deploy, and manage custom AI agents from natural-language prompts.
Architected a Supervisor Agent (LangChain + LangGraph) orchestrating specialized agents — workflow efficiency up 40%.
Designed a visual AI workflow automation platform in the spirit of n8n / Relevance AI.
Implemented real-time Voice AI with LiveKit, OpenAI Realtime/WebRTC, and ElevenLabs — 40% lower end-to-end latency.

LangChainChromaDBLiveKitElevenLabsAzure Service Bus

Led backend engineering and AI architecture for an AI-first scheduling platform serving 200+ automated workflows.
Shipped 50+ production REST APIs (FastAPI) with travel-aware planning — scheduling efficiency up 40%.
Supervisor–worker multi-agent systems reduced manual scheduling conflicts by 60%.
Led a cross-functional team of backend, AI, and frontend engineers.

FastAPILangGraphGoogle Maps APITeam Leadership

Sole backend engineer — built the entire backend of an AI SaaS platform from 0 to production.
Built an AI Meeting Assistant (GPT, Mistral, Llama) automating transcription, summaries, follow-ups, and CRM updates — post-meeting engagement up 60%.
AI agents integrating Slack, Jira, ClickUp, Linear, Gmail, Google Calendar, Zoom, and MS Teams.
Architected AWS + GCP infrastructure at 99.99% availability while cutting infra costs 15%.

Spring BootFlaskNode.jsAWSGCPDocker

03. Skills & Tools

AI & LLM Engineering

Agentic AIMulti-Agent SystemsLangChainLangGraphMCPRAGVector DBsClaude CodeAWS BedrockAzure AIHugging Facen8n

Languages & Frameworks

PythonJavaJavaScriptSQLFastAPISpring BootFlaskDjangoNode.jsReactNext.js

Cloud & DevOps

AWSAzureGCPDockerTerraformTerragruntNginxAzure DevOpsCI/CDMicroservices

Data & Realtime

PostgreSQLMongoDBMySQLRedisChromaDBLiveKitWebRTCElevenLabsAzure Service Bus

04. Things I've Built

sdlc-assistant.ai

$ run sdlc-assistant --env prod

→ spawning multi-agent workforce…

→ RAG index: PDF · DOCX · PPTX · XLSX

✓ user stories generated

✓ manual test cases written

✓ engineering effort −65%

$ ▊

Enterprise · R Systems

AI-Powered SDLC Assistant Platform

Enterprise AI platform that automates the software development life cycle — requirement analysis, user story generation, test case creation, and Azure DevOps work item creation — through a multi-agent architecture with MCP integrations and enterprise RAG.

PythonFastAPILangGraphMCPOpenAIPostgreSQLRedis

agent-platform.ai

$ deploy agent-platform --scale prod

→ supervisor agent orchestrating workers…

→ voice: LiveKit · WebRTC · ElevenLabs

✓ agents built from plain English

✓ voice latency −40%

✓ workflow throughput +45%

$ ▊

Platform · Codewinglet

Enterprise AI Agent Platform

Production-scale platform where users create, deploy, and manage custom AI agents from natural language — Supervisor Agent orchestration, real-time Voice AI, RAG over ChromaDB, and asynchronous processing on Azure Service Bus.

LangChainLangGraphChromaDBLiveKitWebRTCMongoDB

kiroku-scheduler.ai

$ start kiroku-scheduler

→ 200+ automated workflows online…

→ travel-aware planning via Google Maps

✓ 50+ REST APIs shipped

✓ scheduling conflicts −60%

✓ led backend + AI + frontend team

$ ▊

AI-first SaaS · Kiroku Notes

AI Scheduling Platform

AI-first scheduling platform with supervisor–worker agents handling intelligent calendar and event management — travel-aware planning with Google Maps and location intelligence across 200+ automated workflows.

FastAPILangGraphGoogle MapsCalendar APIs

meeting-assistant.ai

$ boot meeting-assistant

→ transcribing meeting audio…

→ GPT · Mistral · Llama pipeline

✓ summaries + follow-ups sent

✓ CRM auto-updated: HubSpot · Salesforce

✓ post-meeting engagement +60%

$ ▊

0 → 1 · Getaligned

AI Meeting Assistant

Built from scratch as sole backend engineer: an assistant that transcribes meetings, writes summaries and follow-ups, handles objections, and updates CRMs automatically — with agents integrating Slack, Jira, Gmail, Zoom, and Microsoft Teams.

Spring BootGPTMistralLlamaAWSGCP

05. Education

Bachelor of Computer Applications (BCA)

Swami Vivekanand Subharti University · Meerut

Graduated Feb 2024

Full Stack Web Development

Masai School · Bengaluru

Graduated Jul 2023

06. What's next?

Let's build something intelligent.

I'm always open to talking about AI platforms, agentic systems, or interesting backend problems — whether that's a role, a collaboration, or just a good technical conversation. My inbox is open.

Say hello

github.com/Amanastel linkedin.com/in/amanastel