Competency:

Portal Automation & MCP Services, Data, Integrations & AI Plumbing, . Backend Service OwnershipJob Mode:

Onsite:

ROLE PURPOSE

RightWalk Foundation works with state governments to make existing entitlements actually reach the people they were written for. ChaturAI is our WhatsApp-based assistant for that job: it walks citizens through NAPS apprenticeship registration and RTE Gujarat admissions, and automates the government portal steps behind the scenes through an AI agent. The backend — a Go service plus Python MCP servers driving live government portals — is the part that has to hold up when a state campaign puts thousands of families on WhatsApp in a single week.

This role owns the server side of ChaturAI: the Go backend that runs the agent loop and conversation state, the Python MCP servers that automate the NAPS and RTE portal journeys, and the PostgreSQL/pgvector data layer underneath. You report to the Engineering Lead, ChaturAI, and work directly with AI/prompt engineers and the programme team in Delhi. This is a high-ownership engineering role, not a ticket-taking one — success depends on building services that stay correct, observable, and recoverable while government portals change without notice.

KEY RESPONSIBILITIES

1. Backend Service Ownership (Go)

- Own the Go backend that receives Chatwoot webhooks, manages the request lifecycle, and persists conversation state in PostgreSQL.
- Drive the agentic orchestration loop — tool-call routing, iteration limits, timeout and error handling — so no user message ever dead-ends silently.
- Design and version the HTTP endpoints and webhook contracts the rest of the stack depends on.
- Hold the quality bar in your area: tests, structured logging, and review of peers' code.

1. Portal Automation & MCP Services (Python)

- Own the Python MCP servers that automate the NAPS and RTE Gujarat portal journeys end to end.
- Build and extend the MCP tool catalogue — new tools, schema changes, and backward compatibility with the agent.
- Lead diagnosis and repair when a government portal changes its markup, flow, or authentication,



with same-day turnaround as the standard.
- Harden automation against flaky upstreams with retries, idempotency, and failure states that surface clearly to the user.

1. Data, Integrations & AI Plumbing

- Own the data model across the portal and bot databases, including migrations and the pgvector collections behind RAG.
- Integrate and operate external surfaces: Meta WhatsApp Cloud API, Chatwoot, and the OpenAI/Bedrock completion APIs.
- Enforce PII discipline — minimise what is stored, and keep beneficiary data out of logs and prompts.
- Maintain the RAG ingestion pipeline and the retrieval quality the agent depends on.

1. Reliability, Observability & Deployment

- Own service reliability for the ChaturAI backend — alerting, escalation, and response during state campaign peaks.
- Drive OpenTelemetry instrumentation into SigNoz so every conversation is traceable end to end.
- Lead incident triage and write the post-incident note yourself.
- Own Docker builds, compose environments, and CI pipelines for your services.

1. Working With the Programme

- Translate requirements from the NAPS and RTE programme teams into technical scope and honest timelines.
- Partner with AI/prompt engineers on guardrails, model routing, and evaluation of agent behaviour.
- Keep technical documentation current enough that a new engineer can run the stack locally on day one.

COMPETENCY MATRIX

Functional (Technical) Competencies

Core (Behavioural) Competencies

Go: production HTTP services, concurrency, and clean webhook/API design.

Ownership: takes a problem to done in production without being chased.

Python: async service work (FastMCP) and resilient automation of third-party web portals.

Debugging temperament:



calm and systematic when a government portal breaks mid-campaign.

PostgreSQL: schema design, migrations, and query tuning; pgvector for retrieval.

Mission fit: motivated by families getting entitlements, not only by clean architecture.

ELIGIBILITY CRITERIA

- Education: B.E./B.Tech or equivalent in Computer Science or a related discipline; strong self-taught engineers with a demonstrable production track record will be considered.
- Experience: 3–5 years building and running production backend services.
- Preferred: exposure to LLM or agentic systems, MCP/tool-calling architectures, WhatsApp or messaging platforms, or Indian government portal integrations.

KEY PERFORMANCE INDICATORS (KPIS)

- Service reliability: uptime and successful-conversation rate for the ChaturAI backend.
- Portal recovery time: median time from a government portal breaking to a verified fix in production.

WHAT SUCCESS LOOKS LIKE — YEAR 1

- Owns the Go backend and both MCP servers independently, shipping to production without supervision.
- Portal breakages are caught by alerting before users report them, with a median fix inside one working day.
- End-to-end tracing in SigNoz covers every conversation, with alerts on the paths that matter.

CANDIDATE PROFILE

- Experience: 3–5 years in backend engineering, with production ownership of at least one live service.
- Robust in Go and Python, and willing to work in both daily — this stack is not single-language.
- Comfortable in a small team with no dedicated DevOps or QA layer: you take your own service to production and keep it up.
- Language: professional English; working Hindi is useful for programme collaboration.

WORK CONDITIONS

- Onsite in New Delhi; occasional travel to programme states. Periodic on-call cover during state campaign peaks.

WHAT WE OFFER / COMPENSATION

- Compensation benchmarked to the Indian social-sector technology market and calibrated to experience — with direct ownership of a production AI system that citizens use to claim government entitlements.

Department:

Technology

📌 Backend Developer (Delhi)
🏢 RightWalk Foundation
📍 Delhi