Sohab Sk

Selected work

AI products shipped to production.

The agentic systems behind my day job: what they do and the scale they run at.

Skillsoft · customer beta

Next-generation agentic authoring for LX Design Studio

Our team originated the idea and owned it end to end, across product direction, engineering, and architecture. I architected it from proof of concept to a production beta with 20+ enterprise customers: multi-agent content generation from uploaded documents, grounded by an in-house vectorless retrieval pipeline built on heuristic document trees and BM25 search.

LangGraphMCPVectorless RAGTypeScript
About LX Design Studio
enterprise beta customers
20+
product, tech, and architecture
Idea → beta
active learners
10M+

Vibe Learning Platform

Codecademy

LangGraph-based agentic engine that generates full-stack, learn-by-building projects, shipped within two months.

LangGraphNext.jsTypeScript
Read the announcement
teams using it
8+

Agentor

Skillsoft internal framework

Reusable framework for multi-agent workflows, MCP tools, guardrails, conversation state, and LLM evaluations. Earned CTO recognition.

TypeScriptMCPLLM evalsGuardrails
pods adopted the model
11

AI-native Builder pod

Skillsoft

Technical lead of Skillsoft's first AI-native pod: three engineers owning delivery end to end, shipping two products in under six months.

Claude CodeAgentic engineeringTechnical leadership

Experience

Production systems, from AI workflows to web platforms.

What shipped in each role, with the scale it reached.

Senior Software Engineer

Skillsoft / Codecademy

Jun 2024 - Present

Technical lead for AI-native product work across Skillsoft and Codecademy: agentic authoring, learning-by-building engines, and the shared frameworks behind them.

  • Originated (with my team) and architected next-generation agentic authoring for LX Design Studio, owning product direction, engineering, and architecture from idea to a production beta with 20+ enterprise customers.
  • Designed its in-house vectorless retrieval pipeline: a non-LLM ingestion step that converts arbitrary document formats into heuristic trees and exposes tree-aware agent tools with in-memory BM25 search, improving grounding and citation fidelity while cutting LLM input tokens on large documents.
  • Shipped Codecademy's Vibe Learning Platform to 10M+ active learners within two months by building a LangGraph-based agentic engine that generates full-stack, learn-by-building projects.
  • Created Agentor, a reusable framework for multi-agent workflows, MCP tools, guardrails, source ingestion, conversation state, and LLM evaluations; reduced agent scaffolding for 8+ teams and earned CTO recognition.
  • Served as technical lead for Skillsoft's first AI-native Builder pod, a three-engineer team that owned product delivery end to end and shipped two products in under six months; the model expanded to 11 additional pods.
  • Built the frontend foundation for LX Design Studio with Next.js, React, TypeScript, reusable routing and design-system patterns, and modular integration across a shared monorepo.
  • Built internal AI developer tooling and release automation, including ACE workflows for sharing reusable skills, subagents, and plugins across 15+ engineering teams.
LangGraphR RAG VectorDB Next.js React TypeScript GraphQL Ruby On Rails GoLang PostgreSQL MongoDBA AWS Redis Kubernetes RabbitMQ

Aug 2022 - Jun 2024

Full-stack blockchain SaaS for EY's OpsChain ecosystem, from smart contracts to Kubernetes delivery.

  • Built full-stack blockchain SaaS products, EY OpsChain Contract Manager and OpsChain ESG, automating contract and sustainability workflows on Ethereum with zero-knowledge privacy components.
  • Developed REST and GraphQL microservices with NestJS and Node.js, containerized with Docker and deployed via Azure Kubernetes Service and API Management.
  • Implemented Starlight zApps and ERC-1155 smart contracts in Solidity, fixed information-security findings, and expanded unit, smoke, and end-to-end coverage with Jest and Supertest.
SolidityV Viem NestJS React GraphQLA Azure Docker Kubernetes Redis MongoDB

Assistant System Engineer

Tata Consultancy Services

Feb 2020 - Aug 2022

MERN-stack product work for Strategy and Transformation, plus Java test automation.

  • Built MERN-stack applications for Strategy and Transformation initiatives using React, Redux Toolkit, Node.js, Express, and MongoDB across core product workflows.
  • Contributed to Java BDD automation frameworks using Selenium and Appium, improving repeatable web and mobile regression testing.
React Redux Node.js ExpressA Azure MongoDBJ Java Python

Projects

Selected public builds and products.

A focused set of shipped tools, libraries, and app projects.

Skills

Focused stack, compact toolbox.

AI systems, product platforms, backend services, delivery, testing, and blockchain work in one tighter view.

LangGraph LangChainM MCPR RAG VectorDBI Ingestion React Next.js SvelteKit Redux Tailwind CSS Node.js NestJS GraphQL Ruby on Rails PostgreSQL MongoDB GoLangA AWSA Azure Docker Kubernetes RabbitMQM Module FederationM Microservices SolidityV Viem Ethereum PolygonZ ZKP Web3.js JavaScript TypeScript Python Ruby Jest Vitest

Additional skills

71 total
L LlamaIndexA A2AM Multi-agent systemsA Agent workflow design Remix Redux Toolkit Material UI shadcn/ui Radix UIG Gamut WebpackR RspackS SWC Vite Rollup ExpressR REST Cloudflare GitHub Actions Jenkins Kafka Apollo FederationM MonorepoD Domain Driven DevelopmentT Test Driven DevelopmentE Event Driven Architecture Ethers.js WagmiH HardhatC ConnectKitR RainbowKitS StarlightJ JavaP Playwright

Ask AI about me

Opens your assistant with my portfolio as context. Pick a question, then a tool.

Question

Gemini can't be prefilled, so its button copies the prompt for you to paste.

Using another AI? Copy this prompt.

Read https://sohab.dev/llms-full.txt (Sohab Sk's portfolio) and summarize who Sohab Sk is, the most notable work shipped, and the strongest areas.

For developers: MCP server and llms.txt

This portfolio is also a read-only MCP server. Add it to Claude Code and your agent can search my work or compare it with a posting.

$claude mcp add --transport http sohab https://sohab.dev/mcp

search_portfolio · compare_with_job_posting · get_experience · get_selected_work · get_projects · get_skills · get_profile · get_contact

Other clients: https://sohab.dev/mcp (Streamable HTTP). Plain text: llms.txt, llms-full.txt.