Projects by Shashank Shekhar - BuildOS, AI, Search & Product Systems

BuildOS Agent

AI-native development workspace for project planning, knowledge management, architecture generation, AI-assisted coding, documentation, and execution.

Problem: Modern AI coding tools lose project context, duplicate decisions, and struggle with long-running multi-repository software projects.

Tech stack: Next.js, React, TypeScript, Tailwind

Outcome: Designed a persistent engineering workspace where project architecture, decisions, docs, implementation status, prompts, and AI conversations become reusable project memory.

BuildOS Knowledge Hub

Knowledge management layer that indexes repositories, docs, architecture notes, prompts, and API specs into structured OKF engineering knowledge for AI agents.

Problem: Engineering knowledge is scattered across repositories, READMEs, docs, prompts, API specs, and architecture notes, forcing developers to repeatedly rebuild context.

Tech stack: Next.js, FastAPI, PostgreSQL, Redis

Outcome: Created a structured project memory layer that extracts applications, modules, APIs, services, schemas, dependencies, decisions, and deployment workflows for AI retrieval.

BuildOS Node Commander

AI-powered infrastructure management platform for self-hosted servers and cloud machines with fleet visibility, Docker operations, remote execution, and migration workflows.

Problem: Managing multiple Linux servers requires switching between SSH, Docker CLI, Portainer, terminal sessions, and monitoring tools.

Tech stack: Next.js, FastAPI, Docker, SSH

Outcome: Built a unified infrastructure workspace for server discovery, Docker inspection, deployments, health monitoring, remote commands, and near-zero downtime Docker migrations.

AI-Integrated Automation Services

Automation services for PDF/text extraction into structured JSON with validation, deterministic post-processing, and fallback paths.

Problem: Product teams needed dependable AI workflow automation without brittle prompts or costly retry loops.

Tech stack: Python, FastAPI, Docker, Ollama

Outcome: Implemented FastAPI-based automation services with deterministic validation, reduced extraction errors, and stable Docker deployment patterns.

AI-Powered Workflow Builder Platform

Visual workflow builder with NL-to-graph generation, versioning, and separation between definition and execution layers.

Problem: Operations teams needed AI workflow automation with clear controls, reusable templates, and auditable execution paths.

Tech stack: FastAPI, SQLite, Alembic, React

Outcome: Shipped a production-ready workflow platform with FastAPI backend services, Docker deployment, and versioned workflow governance.

Data Transformation Adapter

In-progress adapter to transform CSV/JSON into structured templates with export to JSON or CSV.

Tech stack: FastAPI, Python

High-Scale Product Discovery Platform

Enterprise-scale product discovery platform for 7.7M+ SKUs with schema-driven UI, faceted search, and bulk editorial workflows.

Problem: Retail teams had to search and curate millions of SKUs with tools that could not keep up — slow search, no governance, and risky bulk updates.

Tech stack: Angular, Typesense, FastAPI, PostgreSQL

Outcome: Delivered a high-scale Typesense search interface and product engineering workflow that improved discovery speed and editorial consistency.

Homelab / Infra Experimentation

Self-hosted infrastructure experiments to support rapid iteration, reliability, and safe deployments.

Tech stack: Docker, Linux, Proxmox VE, Cloudflare Tunnels

LayerOne Design System

In-progress design system with shared tokens and reusable components across Angular and Next.js apps.

Tech stack: Angular, Next.js, Tailwind, Storybook

Taxonomy & Classification Management System

Governed taxonomy and ML-assisted classification system to manage categories and attributes across enterprise catalogs.

Problem: Catalog teams were classifying millions of products by hand, with inconsistent category rules and no reliable review process.

Tech stack: FastAPI, PostgreSQL, Typesense, Angular

Outcome: Built an Angular-led taxonomy application with ML-assisted review queues that improved classification accuracy and cut manual analyst effort at enterprise scale.