AI Engineer

Convexicon Software Solutions; Gurugram District; On-site; INR 1,800,000 - 3,200,000; Full time

Job summary

Convexicon Software Solutions is hiring an AI Engineer focused on NLP/LLMs and document intelligence to enhanceTender Management capabilities. You will work with the tech lead and product teams to move AI from prototypes to reliable product features used by customers.

A key focus is the Tender Management Solution, helping enterprises understand complex tender documents, identify bid requirements, and reuse credentials to craft stronger bids.

Qualifications

• Strong Python programming and practical experience building AI/ML applications.
• Good understanding of NLP fundamentals and hands-on familiarity with LLMs, prompt engineering, and/or RAG-based applications.
• Experience with at least one major LLM API or model ecosystem, such as OpenAI, Gemini, Anthropic, or open-source Hugging Face models.
• Familiarity with embeddings, vector databases, and semantic search; experience with tools such as FAISS, Qdrant, Pinecone, Milvus, or pgvector is useful.
• Understanding of document processing concepts such as OCR, text extraction, chunking, metadata, retrieval, and structured output validation.
• Working knowledge of ML fundamentals and model evaluation; basic familiarity with computer vision concepts and tools such as OpenCV or YOLO.
• Comfort working with APIs, Git/GitHub, debugging, and collaborating with software engineers to integrate AI features into applications.
• Good to have experience with LangChain, LlamaIndex, Hugging Face Transformers, or similar orchestration/model libraries.
• Hands-on work with scanned PDFs, OCR engines, layout-aware document models, table extraction, or multilingual documents.
• Experience evaluating RAG systems, building test/evaluation datasets, reranking, hybrid search, or observability for LLM applications.

Responsibilities

• Build NLP/LLM pipelines that process lengthy tender documents, including RFPs, RFQs, and eligibility criteria.
• Develop reliable information-extraction workflows to identify and structure key tender fields, with validation and traceability.
• Design and improve Retrieval-Augmented Generation (RAG) systems with grounded answers and citations.
• Work on document ingestion and parsing for PDFs and scanned documents, including OCR and layout-aware extraction.
• Build semantic search, embeddings, reranking, and hybrid retrieval to fetch relevant tender clauses and credentials.
• Develop AI-assisted eligibility checking by mapping tender criteria to company credentials and experience.
• Help convert unstructured documents into a searchable knowledge base for bids and credentials.
• Develop CV and project-experience matching workflows to tender requirements and highlight gaps.
• Create structured outputs such as tender summaries, compliance matrices, risk flags, and submission checklists.
• Evaluate LLM outputs, create test datasets, and implement regression checks to reduce hallucinations.

Skills

Python

NLP

LLMs

CV basics

APIs

LangChain

Education

Bachelor's degree in CS/CE/Math

Tools

FAISS

Qdrant

Pinecone

OpenCV

YOLO

PyTorch

Job description

We are looking for an AI Engineer who is especially interested in Natural Language Processing (NLP), Large Language Models (LLMs), and document intelligence, with working knowledge of Computer Vision (CV). You will work closely with our tech lead and product team to take AI capabilities from prototypes into reliable product features used by real customers. A key focus will be our Tender Management Solution (TMS): helping infrastructure consultancy and other enterprise teams understand complex tender documents, identify bid requirements, assess eligibility, and reuse their company's experience, credentials, and knowledge to prepare stronger bids. This is an opportunity to solve challenging document-understanding and information-retrieval problems, not simply build a generic chatbot.

The core responsibilities for the job include the following:

NLP, LLMs, and Document Intelligence Primary Focus:

• Build NLP/LLM pipelines that process lengthy government and enterprise tender documents, including RFPs, RFQs, eligibility criteria, technical requirements, financial clauses, deadlines, and submission checklists.
• Develop reliable information-extraction workflows to identify and structure key tender fields, with validation, confidence indicators, and traceability to the source page or text.
• Design and improve Retrieval-Augmented Generation (RAG) systems so users can ask questions across tender documents and company knowledge, with answers grounded in retrieved evidence and citations.
• Work on document ingestion and parsing for PDFs and scanned documents, including OCR, layout-aware extraction, tables, multi‑column pages, and document chunking strategies.
• Build semantic search, embeddings, reranking, and hybrid retrieval approaches to retrieve the most relevant tender clauses, past project credentials, CVs, and organizational documents.
• Develop AI‑assisted eligibility checking by comparing tender qualification criteria against company credentials, financial thresholds, certifications, project experience, and personnel requirements.
• Help build the TMS Company Vault: convert unstructured company documents, project sheets, CVs, credentials, and past bid material into a searchable, reusable knowledge base.
• Develop CV and project‑experience matching workflows that map personnel qualifications, roles, years of experience, and project history to tender requirements, while highlighting gaps for human review.
• Create structured outputs such as tender summaries, compliance matrices, risk flags, bid/no‑bid inputs, and submission checklists, with clear evidence for each recommendation.
• Evaluate LLM outputs for extraction accuracy, retrieval quality, groundedness, and consistency; create test datasets and regression checks to reduce hallucinations and brittle prompt behavior.
• Experiment with prompt engineering, tool/function calling, model selection, and, where justified, fine‑tuning or smaller specialized models to balance quality, latency, and cost.
• Build safeguards for missing or conflicting information, low‑confidence outputs, sensitive company data, and human approval of high‑impact bid decisions.

Computer Vision:

• Contribute to computer‑vision use cases across infrastructure monitoring, solid waste management, and operational automation.
• Use OpenCV and detection models such as YOLO to prototype, evaluate, or improve object detection and image/video analysis workflows.
• Understand image annotation, data preparation, augmentation, model evaluation, and deployment considerations for real‑world camera feeds.
• Collaborate with the team to integrate model outputs into production applications and dashboards.

Requirements:

• Strong Python programming skills and practical experience building AI/ML applications.
• Good understanding of NLP fundamentals and hands‑on familiarity with LLMs, prompt engineering, and/or RAG‑based applications.
• Experience with at least one major LLM API or model ecosystem, such as OpenAI, Gemini, Anthropic, or open‑source Hugging Face models.
• Familiarity with embeddings, vector databases, and semantic search; experience with tools such as FAISS, Qdrant, Pinecone, Milvus, or pgvector is useful.
• Understanding of document processing concepts such as OCR, text extraction, chunking, metadata, retrieval, and structured output validation.
• Working knowledge of ML fundamentals and model evaluation; basic familiarity with computer vision concepts and tools such as OpenCV or YOLO.
• Comfort working with APIs, Git/GitHub, debugging, and collaborating with software engineers to integrate AI features into applications.
• Good to have experience with LangChain, LlamaIndex, Hugging Face Transformers, or similar orchestration/model libraries.
• Hands‑on work with scanned PDFs, OCR engines, layout‑aware document models, table extraction, or multilingual documents.
• Experience evaluating RAG systems, building test/evaluation datasets, reranking, hybrid search, or observability for LLM applications.
• Exposure to PyTorch or TensorFlow, YOLO, image augmentation, or real‑time video analytics.
• Experience deploying services to AWS, GCP, Azure, or Linux‑based environments; familiarity with Docker is an advantage.
• Interest in enterprise workflows, bid/tender processes, infrastructure consulting, or document‑heavy business operations.

What We Are Looking For:

• A builder's mindset: you enjoy researching a problem, testing alternatives, and delivering a working solution.
• Product thinking: you care about whether the AI output is accurate, explainable, and genuinely useful to an end user.
• A quality‑first approach: you know that enterprise AI must handle exceptions, cite evidence, and signal uncertainty, not just produce fluent answers.
• Curiosity and ownership: you can communicate your approach, document decisions, and work closely with the tech lead and product team.