AI Solutions Engineer (Forward Deployment Engineer) at Turbham Limited

Role Overview: 

As part of our Solutions Team, you will design and deliver end-to-end AI pipelines tailored to client needs. You will orchestrate our internal services, configuring parameters, apply advanced Prompt Engineering, Data Science principles to validate model performance, while writing the necessary transformation logic to turn raw AI outputs into business-ready data.

Key Responsibilities:

• Pipeline Architecture: Construct end-to-end workflows (ideally, using our proprietary Builder tool), selecting the right AI internal services (or creating custom services), configuring model parameters and implementing custom logic.
• Prompt Engineering & Tuning: Design, test, and refine complex prompts to maximize model performance and ensure high-quality reasoning and results.
• Data Transformation & Logic: Implement Python-based steps to clean, aggregate, and format model outputs. You will apply client-specific business rules to ensure the final payload matches their integration requirements. Also scripts or microservices to bridge feature gaps.
• Evals, Performance Analysis & Optimization: Define KPIs, manage evaluation datasets ("ground truth"), use statistical methods to measure the accuracy and reliability of the solution, and systematically refine prompts until the solution meets accuracy targets.

Requirements:

• LLM Expertise: Deep understanding of LLM capabilities, context management, and Prompt Engineering techniques.
• Data Science Mindset: Experience setting up Evals and data validation, metric definition (Precision/Recall), and analyzing structured vs. unstructured data.
• Solutions Mindset: Ability to translate a client's vague business goal into a deterministic technical pipeline.
• Technical Stack: Strong proficiency in Python for data manipulation, JSON parsing, and logic scripting, including data libraries (e.g., Pandas, NumPy) for manipulation and analysis.

Soft Skills:

• Strong analytical thinking.
• Persistence in iterating solutions (trial and error) to achieve high-quality results.