Lead Applied AI Engineer – Product Intelligence

Employer: Avihs AI Employment Type: Full-time Location: Hybrid/ India Experience: 6–10 years Assignment: Confidential internal land-technology product

ABOUT AVIHS AI

Avihs AI builds intelligent products that convert complex documents, images, operational data, and manual workflows into reliable digital experiences.

We are developing an internal land-technology platform focused on improving property discovery, documentation, verification, monitoring, and related service workflows.

ABOUT THE ROLE

We are looking for a hands-on Lead Applied AI Engineer to establish the AI foundation for this internal product.

You will work directly with founders, product leaders, software engineers, domain experts, verification teams, and field operations. You will be responsible for converting complex land, document, image, location, and operational workflows into dependable product capabilities.

This is not a generic chatbot position. You will help build AI capabilities that become a core part of the product.

The product name and additional business details will be shared with shortlisted candidates during the interview process.

WHAT YOU WILL OWN

- The initial Applied AI architecture for the product. • Feasibility assessment and prioritisation of AI use cases. • Document AI, LLM, Computer Vision, and multimodal pipelines. • AI service APIs and product integrations. • Evaluation datasets and measurable quality standards. • Human-review and exception-management workflows. • Model monitoring, cost controls, and production reliability. • Versioning of models, prompts, datasets, extraction schemas, and validation rules. • Technical guidance for future AI team members.

INITIAL PRODUCT PROBLEMS

You may work on capabilities such as:

- Classification and organisation of land-related documents. • OCR and structured extraction from scanned, photographed, and digital documents. • Extraction of names, plot references, survey details, dates, areas, registration references, and other relevant information. • Comparison of information across multiple documents to identify missing, inconsistent, or conflicting values. • Evidence-linked extraction where every result can be traced back to its source. • Human-assisted document-verification workflows. • Multilingual assistance for English and Indian languages. • Grounded question answering over approved documents, processes, and operational knowledge. • Analysis of property photographs, videos, inspection evidence, and site media. • AI assistance for digital boundaries, location intelligence, and remote property monitoring. • Detection of duplicate, altered, suspicious, anomalous, or low-quality documents and images. • Structured reports for verification, operations, customer support, and management teams.

KEY RESPONSIBILITIES

- Work with product and domain teams to understand property-document, verification, customer, and field workflows. • Convert operational problems into defined AI capabilities, acceptance criteria, and release plans. • Design systems combining OCR, Document AI, LLMs, Computer Vision, structured data, geospatial information, business rules,



and human decisions. • Personally build critical early product components. • Develop secure, testable, and maintainable Python services and production APIs. • Design document-ingestion pipelines covering preprocessing, OCR, classification, extraction, validation, confidence scoring, and evidence capture. • Build grounded knowledge assistants that answer from approved information and identify their supporting sources. • Develop consistency checks and comparisons across multiple documents. • Integrate AI services with backend, web, mobile, mapping, data, and cloud systems. • Create representative datasets and evaluation processes using realistic document and media conditions. • Measure accuracy, coverage, latency, cost, reliability, and human-review effort. • Investigate extraction failures, hallucinations, unfamiliar document formats, poor-quality scans, language issues, and model degradation. • Ensure uncertain or sensitive outputs are routed to authorised human reviewers. • Establish standards for versioning prompts, models, datasets, schemas, validation rules, and evaluation results. • Help recruit and mentor additional AI engineers as the product matures.

RESPONSIBLE AI PRINCIPLE

AI will assist users, employees, and domain experts. It will not independently make unsupported legal, ownership, financing, or government-approval decisions.

The product must clearly distinguish between:

- Information appearing in a source document. • Information extracted by an AI system. • AI-generated observations. • Automated validation results. • Human verification. • Final legal or operational decisions.

MINIMUM REQUIREMENTS

- Six or more years of qualified experience in Applied AI, Machine Learning, Document AI, Computer Vision, NLP, or AI Product Engineering. • Personal ownership of at least two AI systems that reached production or real operational use. • Strong technical depth in Document AI, Computer Vision, Generative AI, or Applied Machine Learning. • Experience building systems combining multiple models, services, APIs, and business workflows. • Strong Python and backend-engineering skills. • Experience creating APIs, pipelines, tests, evaluation systems, monitoring, and production integrations. • Ability to work effectively with incomplete requirements and an evolving product. • Understanding of model limitations, confidence scoring, hallucination risks, traceability, privacy, and human review. • Strong product judgment and the ability to prioritise customer value over technical novelty. • Willingness to remain hands-on while gradually taking technical leadership.

RELEVANT TECHNICAL EXPERIENCE

Experience with a suitable combination of the following will be valuable:

- Python, PyTorch, TensorFlow, Hugging Face, Scikit-learn, or OpenCV. • OCR,



document understanding, layout-aware extraction, and document comparison. • Large Language Models, RAG, embeddings, semantic search, vector databases, structured outputs, and tool calling. • Image classification, object detection, multimodal models, vision transformers, and anomaly detection. • FastAPI, Flask, Django, or comparable backend frameworks. • PostgreSQL, vector databases, object storage, queues, and data-processing pipelines. • Geospatial data, maps, satellite imagery, image change detection, or location services. • Docker, Git, CI/CD, cloud services, logging, monitoring, security, and automated testing. • AWS, Azure, GCP, or comparable cloud platforms.

Direct real-estate or land-industry experience is valuable but not mandatory. Candidates must be willing to learn the domain deeply and work with specialists.

PREFERRED EXPERIENCE

- Building an AI product from prototype through production. • Working with legal, government, financial, construction, insurance, healthcare, or other document-intensive workflows. • Creating evidence-linked AI outputs. • Developing multilingual or India-specific AI capabilities. • Working with geospatial information, satellite imagery, maps, or remote monitoring. • Establishing human-in-the-loop workflows for high-risk decisions. • Building reusable product components instead of project-specific scripts. • Recruiting or mentoring AI engineers. • Working successfully in an early-stage product environment.

WHAT SUCCESS LOOKS LIKE

During the first 90 days, you will:

- Understand the product vision and priority workflows. • Define the initial AI architecture and evaluation standards. • Select one or two high-value use cases for delivery. • Build a working end-to-end document or knowledge-intelligence capability. • Identify required data, annotation, backend, cloud, security, and human-review support.

Within six months, you will:
- Deliver at least one production-quality AI capability. • Establish a representative evaluation dataset. • Implement evidence capture and human review. • Create reusable services supporting multiple document types and regions. • Establish monitoring for quality, latency, cost, failures, and model drift. • Produce a roadmap for multilingual intelligence, property-media analysis, and remote monitoring.

WHAT THIS ROLE IS NOT

- A generic chatbot-development role. • A research-only position. • A management role without hands-on engineering. • A role where complete specifications will always be provided. • A role limited to model training without responsibility for APIs, deployment, users, or business outcomes.

APPLICATION REQUIREMENTS

Please submit

- Your updated résumé. • Your LinkedIn profile. • GitHub, portfolio, publications, research, product demonstrations, or relevant technical work. • Details of two AI systems you personally helped deliver. • Your specific contribution to each system. • One production failure you investigated and resolved. • A brief explanation of how you would extract reliable information from poor-quality scanned documents while preserving source evidence.

📌 Lead Applied AI Engineer – Product Intelligence (Bhubaneswar)
🏢 Avihs
📍 Bhubaneswar