ML Engineer — AI & Operations Research (all genders)
Eorbit Gmbh · Leipzig
Skills in this posting
Extracted from the posting text by the instrument — the demand side, read literally.
The posting
Deine Aufgaben Group and MICE business is one of the last scientifically underserved problems in revenue management. Transient pricing has decades of literature and tooling; group pricing — with its displacement effects, function-space coupling, multi-resource capacity constraints, and negotiation dynamics — remains largely heuristic across the industry. We are building the system that changes that, and this role owns its mathematical core.
You will formulate and solve the optimization problems behind Rocket’s intelligence layer: given a group request, current bookings, forecasted transient demand, and function-space availability — what decision maximizes the hotel’s GOP?
That single objective — measurable gross operating profit uplift for our customers — is the north star of everything you build. These are genuine OR problems: mixed-integer programs, stochastic demand models, displacement cost estimation — and your formulations run in production, pricing real group business for enterprise hotel chains.
Optimization models for group pricing and capacity allocation. Formulate the core decision problems as mathematical programs: MILP formulations for room-block and function-space allocation, displacement-cost models quantifying what a group booking crowds out, and price-recommendation logic with business guardrails as explicit constraints. Own solver strategy (Gurobi / CPLEX / OR-Tools), formulation efficiency, and solution-time guarantees suitable for interactive use.
Demand forecasting and stochastic modeling. Build the forecasting layer the optimizer consumes: transient demand forecasts by segment and stay date, group conversion probability models, cancellation and materialization estimates — with rigorous backtesting on our Databricks data platform.
Insight generation. Build the analytical layer that tells a hotel why — counterfactual analysis (“what would GOP have been under a different pricing policy?”), what-if simulation for revenue managers, and structured recommendations derived from historical RFP and booking data. The output is not a dashboard; it is a defensible, quantified action a revenue director can take.
Scientific rigor in production. Establish the methodological standard: reproducible experiments, benchmark instances, ablations against heuristic baselines, and honest measurement of realized revenue impact at customers. What ships must be defensible — to a hotel’s revenue director and to a referee.
What success looks like
3 months: A first optimization model (e.g., displacement-based price floors) validated against historical booking data and benchmarked against current heuristic practice.
12 months: The optimization core prices group business in production at multiple customers with measured GOP uplift, the insight layer is a selling point in our enterprise deals — and you are positioned to take ownership of the AI team. A body of results strong enough for an OR or revenue-management venue (e.g., INFORMS) is a welcome side effect.
Dein Profil MSc or PhD in Operations Research, applied mathematics, industrial engineering, computer science, or a comparable quantitative field
Solid grounding in mathematical optimization: LP/MILP formulation and solution techniques, and ideally stochastic or robust optimization
Hands-on experience with at least one industrial solver (Gurobi, CPLEX, OR-Tools, HiGHS) beyond coursework
Strong Python and the engineering discipline to ship models as maintainable production services, not notebooks
tatistical modeling and forecasting competence (time series, probabilistic models, backtesting)
Ability to translate a messy business problem into a well-posed formulation — and to explain the solution to a non-mathematical stakeholder
Strong English (B2–C1), German (min. B2)
Nice to have, not required: Background in revenue management, dynamic pricing, or network RM (familiarity with the Talluri & van Ryzin canon)
Publications, or a thesis, in optimization, RM, or applied probability
Experience with decision-focused or learning-augmented optimization
Interest in growing into technical leadership of an AI team
Column generation, decomposition methods, or large-scale MILP experience
Databricks / Spark exposure
Hospitality, airline, or transportation domain experience
Warum wir? Wir sind das Technologieunternehmen, das den Gruppen- und Veranstaltungsverkauf in der Hotellerie neu denkt. Was heute über Posteingang, Excel und Telefon läuft, macht unsere Plattform Rocket zu einem durchgängigen, KI-gestützten Prozess — von der ersten Anfrage bis zum unterschriebenen Vertrag. Europäische Hotelgruppen führen ihren Gruppenverkauf bereits darüber.
Das bedeutet für dich: Du arbeitest nicht trotz KI, sondern mit ihr. Wer bei uns anfängt, gibt Feedback, das in Produktentscheidungen einfließt — und versteht irgendwann, warum ein Prompt besser funktioniert als ein anderer.
Und weil wir schnell wachsen, ist vieles bei uns noch im Entstehen. Manche Antwort lautet ehrlicherweise „das bauen wir gerade noch”. Genau darin liegt die Chance: Du triffst auf Freiräume statt auf fertige Strukturen, und was du hier aufbaust, trägt deine Handschrift.
Was wir bieten Arbeit mit echter KI-Technologie auf internationalem Niveau — nicht als Pilotprojekt, sondern als Geschäftsmodell.
Echter Gestaltungsspielraum. Deine Ideen landen nicht in einem Ideenbriefkasten, sondern in der nächsten Iteration.
Kurze Wege zur Geschäftsführung. Entscheidungen fallen in Tagen, nicht in Gremienschleifen.
Wachstum, das du mitgehst. Wir bauen gerade die nächste Unternehmensphase — mit Rollen, Verantwortung und Perspektiven, die es vor einem Jahr noch nicht gab.
Hauptarbeitsort Leipzig mit regelmäßiger Büropräsenz und flexiblem Homeoffice-Anteil. Hybride Modelle sind bei uns gelebte Praxis, nicht die Ausnahme.
Unbefristeter Vertrag, Vollzeit.
Leistungsprämien für nachweisbare Erfolge.
Mehrtägige Teamevents an Orten, die tatsächlich Spaß machen.
So läuft unser Bewerbungsprozess Intro-Interview — wir lernen uns kennen, du erfährst, wo wir stehen und wohin wir wollen.
Case Study oder Technical Study , je nach Rolle — ein echtes Problem aus unserem Alltag, das wir gemeinsam durchgehen.
Interview mit deiner künftigen Führungskraft und, je nach Position, mit unseren Gründern.
Vom ersten Gespräch bis zur Entscheidung vergehen bei uns maximal vier Wochen. Und du hörst in jedem Fall von uns — auch dann, wenn es diesmal nicht passt.
Find Jobs in Germany on Arbeitnow
Excerpt from the original listing. The full, current text lives at the source. Read and apply there →
The PivotHop read
- What a machine learning engineer actually earnsmedian, seniority, by country
- Careers a machine learning engineer can move intoevery measured route out
- Data Scientist → Machine Learning Engineer52% readiness
- All open machine learning engineer rolesthe full board
Where these skills also reach
Adjacent occupations measured from the same postings — readiness is what a machine learning engineer’s profile already covers.
- 78 open research scientist roles70% readiness from machine learning engineer
- 250 open data scientist roles67% readiness from machine learning engineer
- 146 open ai engineer roles60% readiness from machine learning engineer
- 13 open data annotator roles53% readiness from machine learning engineer
More machine learning engineer roles
- Staff Machine Learning EngineerSailPoint · United States
- Staff Machine Learning EngineerBjak · China
- Staff Machine Learning Engineer, 3D - Singapore Efficiency TeamRiotgames · Singapore
- Senior Machine Learning Engineer (m/f/d)Quantum-Systems GmbH · Gilching
- Senior Machine Learning EngineerIterable · Hybrid - Lisbon, Portugal
Backfilled listing, refreshed with the nightly scrape; the employer has not claimed it yet. Are you the employer? Claim this listing and it can be featured to the candidates whose skills already reach it, first month free.