Generative AI / Machine Learning Engineer

Client

Our client is a leading travel company undertaking a major technology transformation to decouple the frontend user experience from its legacy backend booking engine, enabling greater agility, ownership of mission-critical technologies, and enhanced responsiveness in a fast-evolving travel industry.

Position overview

We are seeking an experienced and innovative Generative AI / Machine Learning Engineer. In this role, the person will serve as a key technical leader in the development of the Itinerary Intelligence Microservice (IIMS)—a platform designed to improve travel content automation for our client.The engineer will be responsible for designing, developing, and deploying a suite of AI‑powered microservices that parse, enrich, compare, and visualize complex travel itineraries. The role will also contribute to the design and implementation of an automated, itinerary‑driven video‑generation pipeline that produces short, high‑quality promotional videos from structured travel itineraries.This includes working with AI‑assisted media selection, asset enrichment and reuse, video‑assembly workflows, and cloud‑based rendering solutions, with a strong focus on scalability, cost efficiency, and production‑ready output.This is a hands‑on position requiring deep expertise in Python, cloud services (preferably AWS), LLM‑based application development, and DevOps practices. The engineer will work on a high‑priority MVP aimed at transforming unstructured data from various sources into valuable, structured assets.

Responsibilities

• Develop and Optimize LLM‑Powered Services: Design, build, and maintain the core Python microservices for the IIMS platform, including:

• Extraction Service: Implement and refine the pipeline that uses vision‑capable LLMs (e.g., GPT‑4 series) to extract structured data from PDF, DOCX, and TXT files.

• Enrichment Service: Engineer prompts and logic to enhance extracted content, generating compelling, market‑ready descriptions for travel components.

• Comparison Service: Build functionality to programmatically compare two document versions and generate a structured summary of differences.

• Disambiguation Service: Create logic to match extracted components (hotels, tours) against a PostgreSQL catalog using fuzzy search techniques and LLM‑driven decision‑making.

• Prompt Engineering & Model Validation: Lead research spikes to test, validate, and iteratively improve prompt performance and select optimal LLMs for each task—balancing cost, speed, and accuracy.

• AI‑Driven Video Generation: Contribute to the design and implementation of an automated, itinerary‑based video generation pipeline, including: AI‑assisted asset selection and enrichment, asynchronous rendering workflows, integration with cloud‑based video assembly tools (FFmpeg, MediaConvert, or third‑party APIs).

• API & Event Interface Design: Design clear, robust service interfaces based on asynchronous messaging (SQS/SNS) and event‑driven patterns.

• DevOps & Deployment: Package Lambda functions using ZIP or container images (Docker where appropriate) to ensure consistent builds and dependency management. Support CI/CD pipelines and infrastructure automation in AWS.

• System Architecture: Collaborate on the design of the end‑to‑end processing pipeline, ensuring services are resilient, scalable, and efficient.

• Prototyping & Technical PoCs: Rapidly build and test proofs of concept for new features, such as the AI‑powered video generation service, including evaluation of external APIs and integration strategies.

Requirements

• Proven hands‑on experience developing complex applications and services in Python. Proficiency with modern Python (3.10+) and asynchronous capabilities (asyncio).

• Demonstrable experience building applications integrated with Large Language Models.

• Deep understanding of prompt‑engineering principles.

• Experience with common LLM APIs (Chat Completions, Assistants/Files APIs).

• Experience with Amazon Bedrock or Amazon Nova.

• Familiarity with handling and processing unstructured data (PDF, DOCX, images, text).