AI Developer – For Chennai location

About the Role CogNet HR Solutions is seeking an AI Developer to design and build our AI-powered intelligent document processing and workflow automation platform which will process complex business documents such as PDFs, scanned documents, spreadsheets, forms, invoices, payroll reports, insurance documents, loss runs, applications, contracts, and other structured and unstructured data. The AI Developer will work across the complete AI pipeline from document ingestion and classification through extraction, validation, reasoning, workflow automation, human review, and enterprise integration. This is a hands-on engineering role for someone who can take AI solutions from prototype to secure, reliable production. Preferred Technology Stack AI / LLM: OpenAI APIs, Azure OpenAI, Amazon Bedrock, Anthropic Claude, Google Gemini, Hugging Face, vision-language models. AI Frameworks: LangChain, LangGraph, Semantic Kernel, LlamaIndex, and agent/tool-calling frameworks. Programming: Strong Python proficiency; FastAPI, Pydantic, REST APIs, JSON and JSON Schema. Experience integrating AI services with C# / ASP.NET Core is highly desirable. Document Processing: PyMuPDF/PDF libraries, OCR technologies, AWS Textract, Azure Document Intelligence, Document AI, OpenCV, and Excel processing. Data: SQL Server/PostgreSQL, vector databases, embeddings, semantic search, Redis or equivalent caching. Cloud / DevOps: AWS, Docker, Git, CI/CD, S3, Lambda, ECS/EKS, CloudWatch, SQS or equivalent messaging/queue technologies.

Requirements

Key

Responsibilities AI & LLM Development

• Design and develop AI applications using Large Language Models (LLMs).
• Build AI agents for document classification, extraction, validation, reasoning, and workflow execution.
• Develop structured extraction pipelines that convert unstructured documents into validated JSON and database records.
• Implement prompt engineering, structured outputs, function/tool calling, and agent workflows.
• Evaluate models based on accuracy, latency, security, and cost.
• Build confidence scoring and human-review routing for exceptions.
• Implement guardrails and validation to minimize hallucinations and incorrect extraction. Intelligent Document Processing Build document-processing pipelines supporting: PDF / Images / Excel / Word / Email → Classification → OCR/Vision → AI Extraction → Validation → Business Rules → Human Review → API/Database.
• Document type detection and classification.
• OCR and vision-based document understanding.
• Table, form, key-value, and entity extraction.
• Multi-page processing, document splitting, and routing.
• Cross-document comparison and reconciliation.
• Confidence scoring, exception handling, and human-in-the-loop review.
• Structured JSON output generation. Agentic Workflow Development
• Develop agents that identify document types and select appropriate extraction schemas.
• Extract and validate information against business rules.
• Compare multiple documents and identify missing or inconsistent information.
• Call APIs and enterprise systems and route documents to appropriate workflows.
• Generate summaries and recommendations and escalate exceptions for human review.
• Design reliable orchestration with retries, logging, fallbacks, and error handling. Application & API Development
• Develop backend AI services and REST APIs.
• Integrate AI services with CogNet applications and third-party enterprise platforms.
• Build asynchronous/background processing for large document workloads.
• Design scalable queues and batch-processing capabilities.
• Implement secure API authentication and authorization. Cloud & Production AI
• Deploy and operate AI applications on AWS.
• Build containerized applications using Docker.
• Implement scalable AI inference and document-processing architectures.
• Monitor model performance, latency, token consumption, failures, and cost.
• Implement production logging, monitoring, retries, and exception handling. RAG & Enterprise Search
• Retrieval-Augmented Generation (RAG), embeddings, vector search, and hybrid search.
• Chunking strategies, metadata filtering, document retrieval, and citation/source grounding.
• Enterprise document search or knowledge-management experience is an advantage. AI Quality & Evaluation
• Create AI evaluation datasets and measure extraction accuracy.
• Compare model performance and track false positives/false negatives.
• Test prompts and extraction schemas and implement AI regression testing.
• Monitor production AI quality and optimize model selection for accuracy, speed, and cost. Security & Data Privacy
• Encryption in transit and at rest, role-based access control, and secure API design.
• Secrets management, audit logging, data isolation, and PII handling.
• Secure LLM integration and protection against prompt injection and malicious documents. Required

Qualifications

• Bachelor's or Master's degree in Computer Science, Artificial Intelligence, Data Science, Engineering, or related discipline.
• 5+ years of software development experience.
• Strong Python programming skills and practical experience building applications using LLM APIs.
• Experience developing REST APIs and backend services.
• Experience processing PDFs and other business documents.
• Strong understanding of structured and unstructured data and SQL databases.
• Familiarity with cloud platforms, preferably AWS.
• Strong analytical and problem-solving skills.

Preferred

Qualifications

• Intelligent Document Processing (IDP) and AI agents/agentic workflows.
• Insurance or Workers’ Compensation documents, payroll/HR data, or financial documents.
• Invoice processing, document reconciliation, and enterprise workflow automation.
• RAG applications, AWS Bedrock, AWS Textract, OpenAI APIs, or high-volume document-processing systems. What You Will Build Document Upload → Automatic Classification → AI Extraction → Validation → Human Review → Workflow → Enterprise Integration
• Insurance submission processing and Workers’ Compensation loss-run analysis.
• Payroll and HR document extraction.
• Benefits reconciliation, invoice processing, and bank reconciliation.
• Contract/form extraction, document comparison, and AI-powered enterprise document search. What Success Looks Like
• Build production-ready document AI pipelines supporting multiple document types and configurable schemas.
• Achieve measurable extraction accuracy and establish automated evaluation and monitoring.
• Develop reusable AI agents and services and integrate Drive with CogNet applications.
• Implement human-in-the-loop exception workflows.

• Improve document-processing speed and automation rates while reducing inference cost. Ideal Candidate We are looking for a builder rather than someone focused only on AI research. The ideal candidate can take a business problem such as: “Read these 200-page insurance documents, identify the document types, extract the required information, validate the information, identify exceptions, and return structured data through an API,” and design and implement the complete production solution. The candidate should be comfortable experimenting with new AI models while maintaining the engineering discipline required for secure, reliable enterprise software.