Job at a glance
About the job AI Backend / Full-Stack Developer – Multilingual AI Assistant Platform Location: Remote; Employment Type: Full-time / Contract; Experience: 3+ Years; Tech Stack: Node.js, TypeScript, LLMs, Qwen, PostgreSQL, Redis, React, Docker About the Role We are looking for an experienced AI Backend / Full-Stack Developer to build and scale a production-ready, multilingual AI assistant capable of understanding English, Hindi, and Hinglish and translating natural-language requests into secure application workflows.
The platform will support account management, wallet operations, payment workflows, transaction history, real-time information retrieval, application navigation, and customer support. An initial Node.js implementation already includes intent classification, structured entity extraction, local Qwen3 inference, a testing interface, and automated tests. You will extend this foundation into a reliable, scalable platform with authenticated API integrations, conversation management, secure transaction handling, monitoring, and production deployment.
This is a hands-on engineering role focused on developing, integrating, testing, and shipping production software. Key Responsibilities Design, develop, and maintain scalable backend services using Node.js and TypeScript. Integrate self-hosted LLMs using OpenAI-compatible APIs or equivalent interfaces. Build multilingual intent recognition and entity extraction for English, Hindi, and Hinglish. Implement conversational context, clarification flows, and multi-step interactions.
Integrate the AI assistant with account, wallet, payment, transaction, and customer support APIs. Develop a secure tool/function-calling framework with schema validation and deterministic business logic. Implement authentication, authorization, transaction confirmation, and audit logging. Handle API failures, retries, timeouts, idempotency, and duplicate requests. Build responsive conversational interfaces and internal management dashboards.
Optimize model inference performance, response latency, and resource utilization. Develop automated evaluation pipelines for intent accuracy, entity extraction, and workflow completion. Write unit, integration, and end-to-end tests. Set up CI/CD pipelines, logging, monitoring, and production deployments. Collaborate with product, frontend, and backend teams to define API contracts and deliver features.
Required Skills 3+ years of software development experience , preferably in backend or full-stack development. Strong proficiency in JavaScript, TypeScript, and Node.js. Experience building and integrating REST APIs. Practical experience integrating LLMs into real-world applications. Understanding of structured LLM outputs, JSON Schema, function calling, and response validation. Experience with authentication, role-based access control, and secure API communication.
Understanding of transactional workflows, state management, retries, and idempotency. Experience with PostgreSQL or similar relational databases. Familiarity with Redis for caching and session management. Working knowledge of Docker, Git, CI/CD, and Linux environments. Experience writing automated tests and debugging production applications. Ability to develop modular, maintainable, and well-documented code.
Ability to work with Hindi and Hinglish inputs, including romanized Hindi and common abbreviations. Good to Have Experience with Qwen, Llama, or other open-weight language models . Familiarity with LM Studio, Ollama, vLLM, or llama.cpp. Experience with React or Next.js. Experience integrating payment gateways, digital wallets, or financial APIs. Knowledge of WebSockets, Server-Sent Events (SSE), message queues, and asynchronous processing.
Familiarity with prompt-injection prevention, sensitive-data redaction, and LLM observability. Experience optimizing GPU utilization, inference speed, and model-serving costs. Understanding of Retrieval-Augmented Generation (RAG). Experience deploying and operating self-hosted AI services. What You'll Work On 1. AI & Natural Language Understanding Improve multilingual intent recognition, entity extraction, conversational context, and structured LLM responses.
2. Backend & API Integrations Connect AI-generated intents to authenticated application APIs with strong validation, permission checks, and error handling. 3. Conversational Workflows Build multi-step conversations, clarification mechanisms, user confirmations, and reliable workflow execution. 4. AI Model Integration & Optimization Integrate and optimize self-hosted models, improve inference performance, and establish evaluation benchmarks.
5. Production Readiness Implement automated testing, monitoring, logging, containerized deployments, and operational documentation. Initial Deliverables Review the existing codebase and recommend a production-ready architecture. Implement authenticated API integration, session handling, and conversation state management. Deliver informational workflows such as account balance, transaction history, and real-time status queries.
Implement secure transaction preparation, validation, and user confirmation flows. Improve multilingual intent accuracy and establish measurable evaluation benchmarks. Deploy the platform to staging with automated tests and monitoring. Support user acceptance testing and production deployment. What Success Looks Like Users can interact naturally in English, Hindi, and Hinglish. The assistant accurately understands requests and extracts relevant information.
Ambiguous requests trigger meaningful clarification. Responses and transaction results are verified against authoritative backend systems. Sensitive operations require authentication, authorization, and explicit user confirmation. API failures are handled gracefully without duplicate transactions or misleading responses. The platform is scalable, maintainable, observable, and ready for future integrations.
How to Apply Your résumé or professional profile. Examples of backend or full-stack applications you have developed. At least one LLM integration project , including your role and technical contributions. A brief explanation of how you would securely connect an AI assistant to APIs that perform financial or account-related operations. We're looking for developers who can build and ship real AI-powered applications, not just chatbot prototypes.