Principal AI/ML Engineer
Information Technology
Europe (Remote)
Remote
Full-time
Senior
5+ years
Posted Aug 18, 2026
About the role
A client of byteSpark.ai is seeking a Principal Applied AI Engineer to provide senior technical direction and hands-on expertise across speech, NLP, large language models and modern applied AI capabilities within secure, high-volume production environments. The role will focus initially on improving speech-to-text quality and reliability, including the evaluation, adaptation and fine-tuning of ASR models such as Whisper or equivalent technologies, while also supporting NLP-driven transcript analysis and the development of AI assistant capabilities using current agentic AI and MCP-based integration approaches. The successful candidate will work closely with existing architects, developers and platform teams to guide model selection, solution architecture, evaluation frameworks, inference optimisation and infrastructure dimensioning, with particular emphasis on on-premises, offline and air-gapped deployment. This is a hands-on principal-level individual contributor role for someone who can combine deep technical judgement with practical model work, mentor engineering teams, and help turn complex AI requirements into reliable production capabilities.
Requirements
- Hands-on production experience in applied AI or machine learning, with strong depth in speech, NLP or related model-driven systems.
- Strong Python capability with practical experience using PyTorch and the Hugging Face ecosystem or comparable modern deep-learning frameworks.
- Production experience with automatic speech recognition and speech-processing pipelines using Whisper, WhisperX, Wav2Vec2 or equivalent technologies.
- Demonstrated ability to adapt, fine-tune, evaluate and improve model performance using measurable quality metrics such as word error rate, latency, throughput or resource efficiency.
- Strong understanding of transformer architectures, LLMs, embeddings, tokenisation, sequence modelling and NLP techniques relevant to classification, entity extraction and conversational AI.
- Practical exposure to modern Agentic AI approaches, tool integration, API or database actions, with current hands-on understanding of MCP or closely related integration patterns.
- Experience deploying or operating open-source AI models in on-premises, offline, self-hosted or air-gapped environments.
- Ability to evaluate model size, hardware requirements and inference trade-offs, with practical exposure to optimisation techniques such as quantisation, ONNX, TensorRT, batching, pruning or distillation.
- Strong technical judgement with the ability to guide architects and engineering teams on model selection, solution design, evaluation strategy and infrastructure requirements.
- Strong English communication, technical documentation, stakeholder collaboration, design review and mentoring capability.
Desirable skills
- Hands-on experience with WhisperX or equivalent advanced ASR frameworks in production environments.
- Experience with multilingual speech recognition, particularly Arabic, regional dialects or other challenging language and accent variations.
- Experience with speaker diarisation, noisy or degraded audio, telephony voice streams, far-field audio or other complex speech-processing conditions.
- Practical experience with real-time or near-real-time speech inference and optimisation for latency, throughput and resource efficiency.
- Experience with NLP techniques for transcript classification, named-entity recognition, sentiment or threat-related analysis using BERT-family models, spaCy or comparable technologies.
- Hands-on experience with quantisation, ONNX, TensorRT or other techniques for optimising models for constrained or specialised hardware.
- Experience delivering AI systems in telecommunications, government, cybersecurity, defence, public-sector or other regulated and security-sensitive environments.
- Experience contributing to AI research, open-source machine-learning projects, technical publications, patents or other evidence of advanced technical depth.
- Exposure to non-generative image or video analytics, multimodal AI or related media-analysis capabilities relevant to future product development.
Role details
- Department
- Information Technology
- Location
- Europe (Remote)
- Work setup
- Remote
- Employment type
- Full-time
- Job level
- Senior
- Experience
- 5+ years
- Industry
- Technology, Information & Media
- Sub-industry
- AI Development
- Compensation
- USD 80,000–90,000 gross annual base + performance bonus + local benefits
Frequently asked questions
This confidential recruitment process is being managed by byteSpark.ai on behalf of an established international technology organisation developing data-intensive products for specialised and security-sensitive environments. byteSpark.ai is not the employer for this position. Further information about the organisation, its products, leadership team, and wider group will be shared with shortlisted candidates at the appropriate stage.
This is an opportunity to work on applied AI problems where model quality, reliability, security, and real-world performance genuinely matter. The successful candidate will help improve production speech-to-text capabilities, contribute to NLP and LLM-based product features, and support the development of modern AI assistant capabilities. The role offers significant technical influence while remaining hands-on with models, experimentation, evaluation, and architecture.
The initial priorities include improving the quality and reliability of existing speech-to-text capabilities and helping establish the technical direction for AI assistant functionality. The successful candidate will work on model selection, adaptation, fine-tuning, evaluation, inference optimisation, and solution architecture while guiding existing engineering teams on how AI capabilities should be integrated into production products.
The organisation operates in highly secure environments where AI capabilities must run on private infrastructure without relying on public AI services or continuous internet access. Candidates should be comfortable working with open-source models, on-premises infrastructure, offline or air-gapped deployment constraints, and the practical trade-offs involved in model selection, performance, compute requirements, and production reliability.
Strong candidates are likely to have hands-on production experience with speech-to-text or Automatic Speech Recognition, particularly model adaptation, fine-tuning, evaluation, or measurable improvements in transcription quality. Experience with Python, PyTorch, Hugging Face, transformer-based NLP or LLM systems, model optimisation, secure on-premises deployment, and practical Agentic AI or tool-integration approaches will also strengthen an application.
The client is looking for a senior hands-on applied AI practitioner who can combine strong technical judgement with practical model development. The successful candidate should be comfortable diagnosing difficult model problems, experimenting directly, making architecture recommendations, and guiding other engineers without needing to operate as a formal people manager.
This is a principal-level individual contributor role with meaningful technical influence across multiple products. The successful candidate will contribute to architecture and model decisions, establish evaluation approaches, recommend infrastructure and model-sizing requirements, review technical designs, and mentor engineering teams. Existing development and platform teams will work alongside the successful candidate on broader implementation and deployment.
This is a full-time remote opportunity focused initially on candidates based in Europe with the appropriate right to work in their country of residence. The gross annual base salary is USD 80,000-90,000, with a performance bonus and locally applicable employment benefits in addition. Employment may be arranged through an existing group entity or an Employer of Record depending on location. Occasional travel to collaborate with teams in Dubai or Kuala Lumpur may be required.