Skip to main content
Shreemanto.dev

Command Palette

Search for a command to run...

Projects

Production systems, AI platforms, and engineering tools. Real problems, not tutorials or clones.

ProductionAI / ML · 2026

Infra AI

Internal AI platform integrating infrastructure systems, operational knowledge, and production tooling for contextual search across 22+ AWS environments.

The Problem

Investigating production issues meant bouncing between dashboards, CLIs, and docs. Context got lost across dozens of AWS environments.

The Solution

Built an internal AI platform that ties infrastructure systems, operational knowledge, and production tooling together so engineers can search and analyze production state in context.

Amazon AgentCoreAWS BedrockMCPPythonKubernetesTerraformSpinnakerGrafanaDatadogPagerDutyAWS
ProductionAI / ML · 2026

The Guardian

Internal agentic AIOps assistant that gives on-call engineers contextual incident help through Slack, Confluence, Amazon Kendra, and platform services.

The Problem

On-call engineers burned time hunting for the right docs and operational context while an incident was already live.

The Solution

Built an internal agentic assistant that pulls operational knowledge and platform context into the incident workflow for faster, more grounded troubleshooting.

Amazon AgentCoreAWS BedrockAmazon KendraSlackConfluenceMCPPython
ProductionAI / ML · 2025

Terraform Plan Intelligence

AWS Bedrock service that summarizes and risk-assesses Terraform plans for 1,000+ release trains/day, including plans with tens of thousands of resource changes.

The Problem

Terraform plans with tens of thousands of resource changes are hard to review quickly when you are shipping at release-train volume.

The Solution

Used semantic chunking and hierarchical summarization so reviewers get deployment summaries and risk signals instead of raw plan noise.

AWS BedrockTerraformPythonSpinnakerAWS
ProductionInfrastructure · 2025

Drift Detection Service

Distributed Go-based cloud-agnostic drift detection across 18+ AWS production cells, cutting infrastructure drift by 70%.

The Problem

Infrastructure drift across many production cells made deployments inconsistent and hard to trust.

The Solution

Designed and shipped a Go-based service that compares Terraform state with live cloud metadata and reduced drift by 70% across 18+ AWS production cells.

GoTerraformSpinnakerAWS
ArchivedBackend · 2024

Online Multiplayer Chess

Real-time multiplayer chess with WebSockets, concurrent sessions, and synchronized gameplay.

The Problem

Needed a clean way to run synchronized two-player chess over the web without clients drifting out of sync.

The Solution

Used Socket.io for live move sync and server-side game logic to manage concurrent sessions.

Node.jsExpressSocket.ioChess.js