The role
We're hiring a Lead AI Engineer to own the intelligence layer inside our proprietary ERP platform, which runs group travel programmes across 23+ European countries.
This is not a research role. You will write production code, own system design, and operate what you build after launch. You'll work directly with IT, operations and sales to find where AI genuinely removes work or improves decisions, then ship it and keep it running.
The role covers two things: agentic systems that perform real operational work, and predictive models that inform commercial decisions.
What you'll do
Agentic AI (primary focus)
- Design and ship production agents for document and email understanding, structured data extraction, and intelligent search across operational systems
- Build multi-agent orchestration: planning and task decomposition, agent memory, state management across steps, and agent-to-agent handoffs
- Define tool schemas and function contracts against existing backend services; enforce structured outputs and constrained decoding where determinism matters
- Own context engineering — RAG pipelines, embeddings, hybrid semantic and keyword retrieval, context window management against cost and accuracy
- Engineer for reliability: retries, idempotency, fallback chains, timeout and partial-failure handling, so agent failures degrade safely rather than corrupt operational data
- Drive model strategy — selection and routing across capability tiers, prompt caching, batching, latency and token cost optimisation at production scale
- Build evaluation frameworks, eval sets and regression suites; measure task completion and extraction accuracy; catch drift when prompts or models change
- Design guardrails, telemetry and human approval gates for AI-initiated actions, with full traceability, aligned to EU AI Act Article 14
Predictive machine learning
- Build forecasting and scoring models for commercial decisions — demand and booking pace by destination, season and source market; inventory utilisation risk; supplier reliability; cancellation probability; quote win-likelihood
- Own the full lifecycle: feature engineering, training, deployment, monitoring, drift detection, retraining
- Know when a model genuinely beats a simple heuristic, and when it doesn't
What we're looking for
- 6–10 years of engineering experience, including 2–3+ years building LLM-backed features that real users depend on daily
- Demonstrable production agentic work: multi-agent orchestration, tool use and function calling, RAG, agent memory, structured extraction, retry and fallback design
- Solid classical ML — regression, gradient boosting, time-series forecasting — with sound judgement on when to apply it
- Strong Python, and willingness to read and integrate with PHP/Laravel services rather than building a separate AI silo
- Hands-on experience with a cloud AI platform — AWS Bedrock preferred; Azure OpenAI, Vertex AI or OpenAI equivalent considered
- Strong SQL and data modelling across relational and document stores
- MLOps fundamentals: deployment, monitoring, drift detection, retraining cadence
- Working discipline around versioning and rollback of prompts, schemas and models; production cost and latency budgets; GDPR-compliant data handling
Communication — specifically:
- You can write a clear technical spec and a one-page decision memo
- You can sit with an operations colleague, understand their real workflow, and turn it into an AI problem statement
- You can explain to a non-technical stakeholder what a model can and cannot be trusted to do, including its failure modes
- You can push back when something should be solved with deterministic code instead of AI
Nice to have
Travel, tourism or hospitality domain exposure. Multilingual NLP. Speech recognition. Experience with MCP or agent developer tooling. Prior mentoring or team leadership.
Pay: ₹1,500,000.00 - ₹2,000,000.00 per year
Work Location: In person
