Fullstack Developer

KEY RESPONSIBILITIES

Design, build, and maintain responsive front-end components using Lit (or equivalent modern component frameworks), including work that crosses Shadow DOM boundaries.

Develop and extend backend services written in server-side ES5 JavaScript, alongside newer TypeScript and Python services.

Build and consume REST APIs, including authentication and role-based access control (Basic Auth, OAuth, JWT).

Model and query relational data, with attention to schema design, joins, and query performance.

Contribute to AI agent services built on LangGraph, including graph design, state management, and checkpointing.

Extend and debug Python/FastAPI services with typed, async code and Pydantic models.

Write and maintain end-to-end and API tests with Playwright (or Cypress/Selenium), including per-role permission checks.

Participate in code review, Git branching/rebasing at team scale, and diagnosing GitHub Actions workflow failures.

Bring up, debug, and operate local environments using Docker Compose.

Use AI coding agents (e.g., Claude Code) responsibly scoping tasks clearly, reviewing generated output critically, and owning the final result.

REQUIRED SKILLS & EXPERIENCE

HRMS Experience is Must

FRONTEND

JavaScript (ES6+) and TypeScript — strong in both.

HTML and CSS/SCSS, including responsive layout and how styling crosses a Shadow DOM boundary.

A modern component framework — Lit ideally, or solid React experience; what matters most is thinking in components, props, and state.

Web components — custom elements, Shadow DOM, reactive properties (direct experience is a plus, not a gate).

Browser fundamentals — ES modules, fetch, DevTools, and basic CSP debugging.

BACKEND

Server-side JavaScript in ES5 — prototype-based classes, no async/await; this is the largest body of code in the codebase.

REST APIs — building and consuming them, with authentication.

Relational data modelling — tables, references, joins, and query performance.

Access control — roles and permissions, Basic Auth / OAuth / JWT; a first-class concern in these applications.

AI AGENT SERVICES

LangGraph — our agent services are built on it: graphs, state, and checkpointing. LangChain or a comparable orchestration library transfers; what matters is having built something agentic, not just called a model.

Python with FastAPI — typed, async, Pydantic models; enough to extend and debug a service, not necessarily own the whole stack.

LLM application basics — prompts, tool calling, structured output, and handling cases where the model returns something unusable.

AWS Bedrock — we call Claude models through it; Anthropic or OpenAI API experience transfers.

TESTING & DELIVERY

Playwright (or Cypress / Selenium) — writing both API and end-to-end tests, including per-role permission checks.

Git at team scale — branches, rebasing, PR review, and resolving real conflicts.

GitHub Actions — enough to read a workflow and diagnose a failure.

Docker Compose — bring up and debug a local stack (up, logs, exec); no Dockerfile authoring needed.

WORKING WITH CLAUDE

Has used an AI coding agent on real work — Claude Code or similar.

Reviews what it writes — can describe a time it was confidently wrong and how they caught it.

Scopes and contextualizes tasks rather than asking for whole features in one go.

Owns the output — can explain and defend every line they submit.

NICE TO HAVE

AngularJS 1.x — our Lit components mount inside AngularJS portal widgets; optional, but rare and hard to teach, so a real bonus. (Not the same as modern Angular.)

ServiceNow — scoped apps, Service Portal, Glide APIs, Fluent SDK.

Python for deployment and provisioning scripts.

Postgres, Redis, OpenSearch; RAG or agent-evaluation experience.