Overview

Location: Chennai — On-site Employment Type: Full-time Working Days: 6 Days a Week Industry: SaaS / Product / AI Technology

Responsibilities

• Design and develop production-ready AI-powered applications using Python.
• Integrate commercial and open-source LLMs into SaaS products.
• Build intelligent AI workflows for document processing, classification, extraction, summarization, question answering, and decision support.
• Develop AI services and APIs that can be integrated with existing web applications.
• Evaluate and select appropriate LLMs based on accuracy, latency, cost, context size, and use case requirements.
• Work with both cloud-hosted and locally hosted/open-source AI models.
• Design and implement Retrieval-Augmented Generation (RAG) pipelines.
• Build document ingestion, parsing, chunking, embedding, indexing, retrieval, and generation workflows.
• Develop semantic and hybrid search capabilities.
• Implement metadata filtering and context-aware retrieval.
• Optimize chunking strategies, retrieval quality, reranking, and context construction.
• Build systems that provide reliable answers grounded in enterprise documents and application data.
• Implement mechanisms to reduce hallucinations and improve response accuracy.
• Work with embedding models for semantic search and document similarity.
• Design and maintain vector search infrastructure.

Qualifications

• Implement document-to-vector and query-to-vector pipelines.
• Optimize vector indexing, similarity search, filtering, and retrieval performance.
• Understand the trade-offs between dense retrieval, sparse retrieval, and hybrid search.
• Implement effective token management strategies for LLM applications.
• Monitor input/output token usage and context-window limitations.
• Optimize prompts and retrieved context to reduce unnecessary token consumption.
• Implement conversation history management and context compression.
• Design strategies for handling long documents and large conversations.
• Optimize AI applications for latency, throughput, reliability, and API cost.
• Implement model fallback and routing strategies where required.
• Develop and maintain structured prompts for production AI workflows.
• Work with system prompts, few-shot prompting, structured outputs, and function/tool calling.
• Build multi-step AI workflows and agent-like systems where appropriate.
• Implement deterministic validation around LLM-generated outputs.
• Design JSON/schema-based AI responses for reliable application integration.
• Continuously evaluate and improve prompts based on real-world results.
• Build AI pipelines for processing resumes, PDFs, documents, forms, and other unstructured data.
• Implement document extraction and normalization workflows.
• Extract structured information from unstructured documents using LLMs and traditional processing techniques.
• Handle OCR-generated and imperfect document data.
• Build document classification, entity extraction, summarization, and enrichment pipelines.
• Design scalable asynchronous processing workflows for large document volumes.
• Develop scalable backend services using Python.
• Build REST APIs and AI service endpoints.
• Work with frameworks such as FastAPI, Flask, or Django.
• Integrate AI services with existing frontend and backend systems.
• Implement authentication, authorization, validation, logging, error handling, and monitoring.
• Design asynchronous and background processing systems for long-running AI jobs.
• Work extensively with PostgreSQL and relational data models.
• Design schemas for AI-related data and application workflows.