Senior AI/Software Expert – HMRC R&D Technical Review

Independent AI / Software Expert – HMRC R&D Technical Review

Project Overview

We are a UK-based recruitment and technology business seeking an experienced software engineering / applied AI professional to undertake an independent technical review supporting an HMRC R&D Corporation Tax relief claim.

We have developed Zawa, a proprietary AI-enabled recruitment platform incorporating large language models (LLMs), candidate scoring, CV extraction and automated recruitment workflows.

Our accountants have requested an independent assessment by a suitably qualified competent professional to establish whether the challenges encountered during development represented genuine technological uncertainties under HMRC's R&D requirements.

Specifically, we need to establish whether these challenges could have been readily resolved by an experienced professional using established software engineering and AI techniques, or whether the development sought an advance in technological capability.

The claim covers development undertaken during the accounting period ended 28 February 2026.

Technical Areas for Review

1. AI Candidate Scoring

Development of AI functionality to evaluate candidates against structured vacancy requirements.

Challenges included:

Achieving consistent scoring from non-deterministic AI models.

Maintaining accuracy across large candidate datasets.

Preventing hallucinated or unsupported assessments.

Grounding scores in actual candidate and vacancy information.

Managing output truncation, incomplete responses and differences between LLM architectures.

2. AI-Driven CV Extraction

Development of functionality to extract structured information from unstructured CV documents.

Challenges included:

Extracting complete employment histories, qualifications and skills from varying formats.

Preventing missing or fabricated information.

Maintaining accurate, consistent structured outputs.

Overcoming limitations of existing LLM extraction techniques.

Developing and testing approaches to improve reliability.

We recognise that using existing AI models or implementing standard software solutions does not automatically qualify as R&D. We require an expert capable of distinguishing routine engineering from genuine technological uncertainty.

Scope of Engagement

We have an existing technical report (approximately 11 pages) prepared with our accountants, documenting the development work, technical problems, testing and solutions.

The successful applicant will:

Review the technical report and available supporting evidence.

Assess the technological baseline and established AI/software capabilities available when development began in August 2025.

Determine whether the identified challenges constituted genuine technological uncertainties that could not readily have been resolved by a competent professional.

Assess whether the development sought an advance in overall technological capability.

Identify technical inaccuracies, unsupported claims or areas requiring further evidence.

Participate in a discussion with our technical development lead.

Produce a concise written assessment with conclusions and recommendations.

Essential Requirement – HMRC Competent Professional

We specifically require someone willing to put their professional name and credentials to an independent technical assessment supporting an HMRC R&D claim.

The successful applicant must be willing to:

Provide their name, qualifications and relevant professional experience.

Confirm their competence in the relevant software engineering and AI disciplines.

Provide a signed written assessment of the technological advances and uncertainties.

Be available for reasonable follow-up clarification if HMRC enquires into the claim, subject to agreed terms.

We are not seeking a guaranteed positive opinion or rubber-stamp approval. The assessment must be independent, objective and technically defensible.

Required Experience

Applicants should have substantial experience in:

Software engineering and architecture.

Applied AI, LLMs and natural language processing.

AI reliability, model evaluation and structured data extraction.

Production AI systems.

Previous experience supporting UK HMRC R&D tax relief claims is highly desirable.

We particularly welcome applications from senior software architects, experienced AI engineers and technical consultants familiar with assessing technological uncertainty.

Deliverables and Budget

We envisage a focused engagement involving:

Review of the technical documentation.

One technical discussion (45–60 minutes).

Concise written assessment identifying potentially qualifying and non-qualifying technical activities.

Recommendations for improving the report.

Signed professional statement setting out the reviewer's conclusions and credentials.

Fixed budget: £750, depending on experience. We are open to proposals from suitably qualified specialists.