About the role

Position: Senior AI Engineer — Technology & Product

Role Overview

The Senior AI Engineer is AMPYR GTC’s hands-on builder for applied AI — writing the production code that turns Large Language Models, Retrieval-Augmented Generation, and domain-adapted models into reliable internal tools for analysts, engineers and decision-makers. You will work directly with the GM — Product & Technology, who owns the AI strategy and architectural direction; your job is to execute, challenge, and ship. You will pair closely with a Full-Stack Developer on the team and collaborate with domain experts across our international platforms. This is a foundational individual-contributor hire — the production systems you build in your first year will shape AMPYR’s AI capability for the next several.

Key Responsibilities

Build production Retrieval-Augmented Generation (RAG) pipelines over enterprise document and data corpora — hands-on, end-to-end Python development covering ingestion, chunking, embedding, retrieval, re-ranking, and evaluation.
Ship domain-adapted LLM capabilities through prompt engineering, fine-tuning, or adapter-tuning techniques, with measurable improvements in task accuracy and cost.
Build evaluation, monitoring, and guardrail systems for LLM-based applications — hallucination detection, citation validation, drift monitoring, and continuous quality measurement.
Implement LLM features inside internal web applications — writing the Python and API code that integrates inference, retrieval, and tool-use into user-facing workflows.
Make and present build-vs-buy recommendations between hosted inference providers and self-hosted models, supported by cost, accuracy, latency, and data residency analysis.
Pair with another engineer on the team through code review, architecture walk-throughs, and joint debugging. This is peer collaboration, not line management.
Work directly with domain experts to translate subject-matter workflows into AI-assisted tooling that demonstrably reduces manual effort and improves decision quality.
Set up MLOps foundations — model versioning, eval pipelines, prompt management, and deployment workflows. You will build these yourself, not specify them for others to build.
Track AI infrastructure costs continuously and surface trade-offs before they become budget surprises.
Contribute hands-on to forward-looking initiatives such as forecasting and optimisation engines as the platform expands.

Qualifications, Experience and Skill Set

Qualification
- Bachelor’s degree in Computer Science, Engineering, Mathematics, Statistics, or a related quantitative field. Master’s degree welcomed but not required.
- Relevant certifications (AWS / Azure AI, MLOps, model deployment) considered favourably.
Years of experience
- 4–6 years total professional experience in software engineering, machine learning, or data science.
- Minimum 2 years of production experience with Large Language Models — building, deploying, and maintaining LLM-based applications.
Must-have skill set
- Strong proficiency in Python and the modern AI / ML ecosystem (LangChain, LlamaIndex, Hugging Face, or equivalents).
- Hands-on experience designing and deploying production RAG systems — chunking strategy, hybrid retrieval, re-ranking, evaluation, and hallucination handling.
- Practical experience with foundation model APIs (OpenAI, Anthropic Claude, Azure OpenAI, AWS Bedrock) and at least one open-weight model family (Llama, Mistral, or similar).
- Experience with vector databases (Chroma, Pinecone, pgvector, Weaviate, or similar) and embedding model selection trade-offs.
- Strong understanding of prompt engineering, function calling / tool use, and agentic patterns.
- Production experience with Python web frameworks (FastAPI or Flask), PostgreSQL, Docker, and AWS (or equivalent cloud).
- Demonstrated ability to make build-vs-buy decisions on AI infrastructure with cost discipline.
- Excellent analytical skills with the ability to interpret complex data and reason about model behaviour, error modes, and failure cases.
Good-to-have skill set
- Experience fine-tuning or adapter-tuning foundation models (LoRA, QLoRA, SFT, instruction tuning).
- Exposure to MLOps tooling (MLflow, Weights & Biases, LangSmith, Langfuse, or similar).
- Experience with document intelligence workflows — OCR, layout-aware parsing, tabular extraction (Docling, Unstructured, Textract).
- Familiarity with the energy, infrastructure, or financial services domain.
- International or cross-cultural collaboration experience — comfort working with stakeholders in the UK, Europe, and the USA across time zones.
- Open-source contributions, technical writing, or conference presentations in the applied AI space.
Communication and leadership
- Strong communication and presentation abilities — able to explain architectural choices and trade-offs clearly to non-technical leadership and domain experts.
- Comfortable pairing with and informally coaching a junior engineer through code review and joint problem-solving; formal people-management experience is not required and is not part of this role.
- Comfort working independently in a small, senior team without a heavy management layer.
- Bias toward shipping working software over producing documentation about software.