Forward-Deployed AI Engineer

Job Description: Forward-Deployed AI Engineer 

About the Team 

Enterprise Data Services (EDS) is Bottomline's AI Engineering function. We don't build AI as a side tool — we build AI-native solutions where intelligence IS the product. Our platform runs on Snowflake, and we use Snowflake's native AI capabilities (Cortex AI, Cortex Agents, Cortex Search, Cortex Analyst, ML Functions) to solve business problems directly in the data platform, no external ML infrastructure, no separate serving layer, no handoffs between teams. 

We currently ship: churn prediction agents, contract intelligence, partner reporting automation, a governance control plane, and much more. Everything runs in Snowflake. Everything is versioned, tested, and production-governed. 

The Role 

We're looking for an Forward-Deployed AI Engineer (FDE) who can take a business problem and solve it natively in Snowflake — from feature engineering through model training to deployment as a live service. This is not a platform-building role. You won't be designing schemas or managing pipelines. You'll be the person who turns business questions into working AI solutions that run in production. 

What You Will Do 

• Take business problems (e.g. churn risk, propensity to buy, anomaly detection, document intelligence) and design AI solutions using Snowflake-native capabilities 

• Build and evaluate ML models using Snowflake ML Functions, Snowpark ML, and Cortex AI 

• Design and implement Cortex Agents that combine structured data queries with unstructured document intelligence 

• Create evaluation frameworks (golden datasets, LLM-as-judge, human review loops) to prove solutions meet success metrics 

• Work with Cortex Analyst semantic views to enable natural language access to business data 

• Build Cortex Search services over unstructured documents (contracts, support tickets, product docs etc.) 

• Collaborate with AI and Data Engineers who handles platform infrastructure, you focus on the science, they handle the plumbing 

• Present findings and recommendations to business stakeholders in plain language 

• Contribute to the AI project lifecycle: framing problems, running spikes, evaluating results, and iterating 

Requirements 

Must Have 

• 3-4 years experience in data science, machine learning, or applied AI 

• Strong Python (pandas, scikit-learn, numpy), you think in DataFrames 

• SQL fluency, comfortable writing complex analytical queries, window functions, CTEs 

• Experience building classification or regression models and evaluating them rigorously (precision, recall, F1, AUC) 

• Experience with at least one of: NLP, document extraction, recommendation systems, or anomaly detection 

• Comfortable working with messy real-world data — missing values, inconsistent formats, multiple sources 

• Great communications skills and can work directly with stakeholders end to end 

• Can communicate findings to non-technical stakeholders clearly 

Strong Preference 

• Snowflake experience (Snowpark Python, Cortex AI functions, ML model registry) 

• Experience with LLMs in production (prompt engineering, RAG architectures, evaluation of generative outputs) 

• Understanding of Cortex Agents, Cortex Search, or Cortex Analyst 

• Experience building evaluation sets and measuring AI quality systematically (not just accuracy, precision/recall tradeoffs, error analysis) 

• Familiarity with feature engineering for tabular business data (customer signals, transaction patterns, contract attributes etc) 

Nice to Have 

• Experience in Fintech, B2B payments, or B2B SaaS 

• Familiarity with A/B testing or causal inference methods 

• Experience with Streamlit for rapid prototyping of data apps 

Our Tech Stack 

• Snowflake (everything lives here): Cortex AI, Cortex Agents, Cortex Search, Cortex Analyst, Snowpark Python, ML Functions, Dynamic Tables 

• Python: pandas, scikit-learn, snowpark-ml, evaluation frameworks 

• Version control: Git, Snowflake stage-based deployment 

• Visualisation: Streamlit in Snowflake, Cortex Code (IDE), PowerBI 

Why This Role Is Different 

• AI-native, not AI-assisted: You will build solutions where the AI IS the product, not a feature bolted onto something else 

• No ML infrastructure tax: Everything runs natively in Snowflake — no Docker, no Kubernetes, no model serving endpoints to manage 

• You will have business impact from day one: with a prioritised backlog of Quick Win candidates waiting for someone to build them 

• Direct line to stakeholders: No layers of middle management between you and the business problem. 

• Modern AI engineering practices: Lifecycle-governed (Framing > Spikes > Build > Evaluate > Governance > Adoption), evaluation-driven, with a real governance plane 

Your First 90 Days 

Days 1-30: Learn 

• Understand the Snowflake platform and Cortex AI capabilities 

• Run through the AI project lifecycle on a small candidate 

Days 30-60: Spike 

• Own your first AI candidate from the prioritised backlog  

• Run a timeboxed spikes: build the crudest version that proves positive signal 

• Create your first golden evaluation dataset 

• Present spike findings to the team 

Days 60-90: Build 

• Build your spike to production standard with proper evaluation 

• Pass through governance gates 

• Ship your first AI solution to production