Deep hands-on experience building production GenAI applications using OpenAI, Anthropic, Gemini, Llama, Mistral, or equivalent. Proficient in orchestration frameworks (LangChain, LlamaIndex, CrewAI, o
RAG Systems
Proven ability to design and tune RAG pipelines: chunking, embedding, vector store selection (Pinecone, Weaviate, pgvector, Chroma), hybrid search, and re-ranking. Understands the failure modes of nai
Prompt Engineering
Systematic approach to prompt design: structured outputs, tool use, chain-of-thought, few-shot selection, and prompt regression testing.
Model Fine-Tuning
Experience with parameter-efficient fine-tuning methods (LoRA, QLoRA), dataset curation for instruction tuning, and evaluation-driven iteration.
Model Evaluation
Ability to design rigorous evaluation frameworks: offline metrics, human preference studies, red-teaming, and A/B testing in production.
Classical ML
Strong foundation in supervised and unsupervised learning, statistical modeling, and time-series analysis. Knows when not to use an LLM.
Python
Expert-level Python. Comfortable with PyTorch or JAX for model work, plus standard data stack: pandas, NumPy, scikit-learn, and SQL.
MLOps
Experience managing the full ML lifecycle in production: experiment tracking, model versioning, CI/CD for ML, monitoring, and cost management.
Cloud & Infrastructure
Practical experience deploying ML workloads on AWS, Azure, or GCP. Comfortable with containers, managed inference endpoints, and GPU provisioning.
Nice to have
Graduate degree in Computer Science, Statistics, Mathematics, or a related quantitative field, or equivalent research and industry experience.
6 or more years of industry experience in data science or applied ML, with at least 2 years focused on LLM or GenAI application development.
Experience in commercial real estate, capital markets, investment management, or financial services.
Published research or public contributions (papers, open-source, technical blog posts) in AI, ML, or NLP.
Familiarity with knowledge graphs, structured reasoning, or neuro-symbolic approaches for grounding model outputs in verified facts.
Experience with multi-modal models (vision-language models, document understanding) for processing financial documents, floor plans, or property imagery.
Exposure to responsible AI practices: bias auditing, fairness metrics, model explainability, and regulatory considerations for AI in financial services.
What You Will Work On
CBRE's GenAI applications span a wide range of domains. Below are examples of the types of systems this role will contribute to -- not an exhaustive list, since the roadmap evolves as we learn.
Application Area
Full description
Senior Data Scientist
GenAI Application Development
CBRE Technology & Digital
September 2026
Role Overview
CBRE is hiring a Senior Data Scientist to lead the design and delivery of generative AI applications that transform how our business operates and how our clients experience our services. This is a senior individual-contributor role with real technical ownership: you will architect solutions, not just advise on them.
You will work at the frontier of applied AI, building production-grade GenAI systems that solve concrete business problems across advisory, asset management, research, and operations. You will partner with product managers, engineers, and domain experts to move ideas from whiteboard to working software -- and you will set the technical bar for how CBRE thinks about and deploys large language models.
Key Responsibilities
GenAI Application Development
Lead end-to-end development of GenAI applications including document intelligence, conversational agents, automated reasoning pipelines, and decision-support tools.
Design and implement retrieval-augmented generation (RAG) architectures, selecting the right chunking strategies, embedding models, vector stores, and retrieval methods for each use case.
Build multi-agent orchestration systems that decompose complex tasks, route between specialized agents, and maintain coherent state across long-running workflows.
Evaluate and select the right foundation models (proprietary and open-source) for each application, balancing capability, cost, latency, and data-residency requirements.
Apply prompt engineering rigorously, using structured outputs, chain-of-thought reasoning, tool use, and few-shot examples to shape model behavior reliably.
Model Development and Fine-Tuning
Fine-tune and adapt foundation models using techniques such as LoRA, QLoRA, instruction tuning, and RLHF where off-the-shelf models fall short.
Design and execute systematic model evaluation frameworks: define metrics, build evaluation datasets, run human preference studies, and track regressions across model versions.
Apply classical ML and statistical methods alongside LLMs where appropriate, including time-series forecasting, clustering, classification, and anomaly detection for structured real estate and financial data.
Stay current with the research literature and rapidly evaluate new techniques -- papers published one week should be prototyped the next.
MLOps and Production Engineering
Senior Data Scientist at CBRE (Noida; Hyderabad; Gurugram) — aijobsearch
Communication
Able to present findings, trade-offs, and recommendations to senior business stakeholders without hiding behind technical jargon. Writes clearly and concisely.
Own the full lifecycle of models you build: from experimentation through staging, production deployment, monitoring, and iteration.
Instrument GenAI applications with appropriate observability -- trace latency, token usage, retrieval quality, and output correctness in production.
Build and maintain experiment tracking, model registries, and evaluation pipelines using tools such as MLflow, Weights & Biases, or equivalent.
Work with platform engineers to deploy models and inference services on cloud infrastructure (AWS, Azure, or GCP), ensuring reliability, scalability, and cost efficiency.
Implement guardrails, content filters, and safety checks that make GenAI applications suitable for enterprise deployment.
Data and Research
Interrogate large, heterogeneous datasets -- structured property data, unstructured documents, market research, financial records -- and extract signal that drives model quality.
Design data pipelines that prepare training, evaluation, and retrieval corpora at scale.
Apply domain knowledge of commercial real estate or financial services to guide feature engineering, evaluation criteria, and solution design.
Publish internal research notes and conduct knowledge-sharing sessions that elevate the broader team's technical understanding.
Stakeholder Collaboration
Translate complex model behavior and trade-offs into language that non-technical stakeholders -- executives, advisors, analysts -- can act on.
Work directly with business lines to identify high-value AI opportunities, scope solutions realistically, and set honest expectations about what models can and cannot do.
Mentor junior data scientists and share expertise across the AI community of practice.
Write clear technical documentation covering system design, model cards, evaluation results, and operational runbooks.
Required Qualifications
Area
What We Are Looking For
GenAI & LLMs
Deep hands-on experience building production GenAI applications using OpenAI, Anthropic, Gemini, Llama, Mistral, or equivalent. Proficient in orchestration frameworks (LangChain, LlamaIndex, CrewAI, or DSPy).
RAG Systems
Proven ability to design and tune RAG pipelines: chunking, embedding, vector store selection (Pinecone, Weaviate, pgvector, Chroma), hybrid search, and re-ranking. Understands the failure modes of naive RAG and knows how to fix them.
Prompt Engineering
Systematic approach to prompt design: structured outputs, tool use, chain-of-thought, few-shot selection, and prompt regression testing.
Model Fine-Tuning
Experience with parameter-efficient fine-tuning methods (LoRA, QLoRA), dataset curation for instruction tuning, and evaluation-driven iteration.
Model Evaluation
Ability to design rigorous evaluation frameworks: offline metrics, human preference studies, red-teaming, and A/B testing in production.
Classical ML
Strong foundation in supervised and unsupervised learning, statistical modeling, and time-series analysis. Knows when not to use an LLM.
Python
Expert-level Python. Comfortable with PyTorch or JAX for model work, plus standard data stack: pandas, NumPy, scikit-learn, and SQL.
MLOps
Experience managing the full ML lifecycle in production: experiment tracking, model versioning, CI/CD for ML, monitoring, and cost management.
Cloud & Infrastructure
Practical experience deploying ML workloads on AWS, Azure, or GCP. Comfortable with containers, managed inference endpoints, and GPU provisioning.
Communication
Able to present findings, trade-offs, and recommendations to senior business stakeholders without hiding behind technical jargon. Writes clearly and concisely.
Preferred Qualifications
Graduate degree in Computer Science, Statistics, Mathematics, or a related quantitative field, or equivalent research and industry experience.
6 or more years of industry experience in data science or applied ML, with at least 2 years focused on LLM or GenAI application development.
Experience in commercial real estate, capital markets, investment management, or financial services.
Published research or public contributions (papers, open-source, technical blog posts) in AI, ML, or NLP.
Familiarity with knowledge graphs, structured reasoning, or neuro-symbolic approaches for grounding model outputs in verified facts.
Experience with multi-modal models (vision-language models, document understanding) for processing financial documents, floor plans, or property imagery.
Exposure to responsible AI practices: bias auditing, fairness metrics, model explainability, and regulatory considerations for AI in financial services.
What You Will Work On
CBRE's GenAI applications span a wide range of domains. Below are examples of the types of systems this role will contribute to -- not an exhaustive list, since the roadmap evolves as we learn.
Application Area
Description
Document Intelligence
Automated extraction, classification, and synthesis of lease abstracts, appraisal reports, due diligence packages, and research publications.
Conversational Research
AI research assistants that let advisors and analysts query CBRE's proprietary data, market reports, and benchmarks in natural language.
Workflow Automation
GenAI-powered pipelines that automate repetitive analytical tasks -- market comps, portfolio summaries, client briefing drafts -- with human-in-the-loop review at critical decision points.
Decision Support
Models that surface relevant insights at key moments in the advisory and transaction process, helping brokers and advisors act faster with better information.
Structured Data Reasoning
Systems that combine LLMs with structured financial and property data to answer complex questions that neither approach could handle alone.
What Success Looks Like
In your first 90 days, you will have shipped a meaningful improvement to an existing GenAI system -- better retrieval quality, reduced hallucination rate, lower latency, or a new capability -- with the metrics to prove it. Within six months, you will have designed and launched at least one new GenAI application from the ground up and built a clear point of view on where CBRE's AI capabilities need to go next. Your work will be in production, not in a notebook.
About CBRE
CBRE Group, Inc. is the world's largest commercial real estate services and investment firm, with services, insights, and data that span every dimension of the industry. We create solutions for clients of every size, in every sector, and across more than 100 countries. CBRE is an equal opportunity employer.
CBRE 2026 All Rights Reserved. All information included in this proposal pertaining to CBRE—including but not limited to its operations, employees, technology and clients—are proprietary and confidential, and are supplied with the understanding that they will be held in confidence and not disclosed to third parties without the prior written consent of CBRE. This letter/proposal is intended solely as a preliminary expression of general intentions and is to be used for discussion purposes only. The parties intend that neither shall have any contractual obligations to the other with respect to the matters referred herein unless and until a definitive agreement has been fully executed and delivered by the parties. The parties agree that this letter/proposal is not intended to create any agreement or obligation by either party to negotiate a definitive lease/purchase and sale agreement and imposes no duty whatsoever on either party to continue negotiations, including without limitation any obligation to negotiate in good faith or in any way other than at arm’s length. Prior to delivery of a definitive executed agreement, and without any liability to the other party, either party may (1) propose different terms from those summarized herein, (2) enter into negotiations with other parties and/or (3) unilaterally terminate all negotiations with the other party hereto. CBRE and the CBRE logo are service marks of CBRE, Inc. All other marks displayed on this document are the property of their respective owners, and the use of such logos does not imply any affiliation with or endorsement of CBRE.