New opportunity

AI Engineer, Associate – Data Analytics & AI

Blackstone · New York, United States of America

About this role

Blackstone seeks an AI Engineer for BXTI’s Alternative Asset Management Technology Data Analytics & AI team, supporting BXMA, Private Equity, Real Estate, and Private Wealth. The role aims to accelerate AI adoption by working with business teams to identify use cases and deliver solutions that fit their workflows.

Responsibilities

Embed with business units, shadow live workflows, engage AI users, and learn how teams raise, service, and report on capital. Turn workflow friction into a prioritized pipeline of self-service improvements and larger platform initiatives, validating demand before building.
Develop and maintain scalable, reliable, secure production software using Python and SQL. Apply version control, CI/CD, containerization, and cloud deployment practices, including tools such as Git, Docker, and Terraform.
Build and extend agentic workflows, retrieval pipelines, and evaluation tooling. Release validated self-service runbooks and prompt libraries to support adoption.
Lead multiple concurrent projects, move proofs of concept into supported production deployments, and deliver talks and training for technology and business teams.

Qualifications

3+ years in AI engineering, software engineering, or data analysis; fluency in Python and SQL for working with data and data warehouses.
Applied experience with LLMs and AI tooling, such as fine-tuning, prompt engineering, agentic workflows, retrieval, or model evaluation and benchmarking. Experience shipping and maintaining systems under real user load, using software development lifecycle tools, and learning unfamiliar technology with minimal supervision.
Experience working with non-technical stakeholders to understand processes and shape solutions; experience on a technology team at a financial firm; strong communication, documentation, project scoping, and prioritization skills.
A bachelor’s or master’s degree in Computer Science, Engineering, Finance, or Economics. Experience in alternative asset management, private wealth, or fund distribution is preferred.

Compensation and benefits: Expected annual base salary is $110,000–$170,000. Benefits include health coverage, paid time off, life insurance, a 401(k) plan, and discretionary bonuses; some employees may be eligible for equity or other incentives. Depending on duties, the role may require securities licenses.

For full details about this role, click “Apply on company site” to view the employer’s original posting.

Skills for this role

PythonSQLLarge Language ModelsFine-tuningPrompt EngineeringAI AgentsRetrievalModel EvaluationBenchmarkingGitCI/CDDockerTerraformContainerizationCloud DeploymentData WarehousesStakeholder EngagementRequirements GatheringTechnical DocumentationProject Scoping

Checking your profile…

Preparation suggestions, not the employer’s actual interview questions. Always check the original posting for current requirements.

Understand the role

What matters most

Discover and validate workflow needs
The role embeds with BXMA, Private Equity, Real Estate, and Private Wealth teams to shadow how they raise, service, and report on capital. It calls for validating demand across teams before building.
Prioritize improvements by scope
Observed workflow friction must become a pipeline of self-service quick wins and larger platform initiatives. The posting also emphasizes defining scope and communicating changes.
Ship maintainable production software
The engineer develops Python and SQL solutions with attention to quality, scalability, reliability, and security, using version control, CI/CD, containerization, and cloud deployment practices. The posting distinguishes maintained systems under real user load from prototypes.
Build and evaluate applied AI
The work includes agentic workflows, retrieval pipelines, and evaluation tooling. Applied LLM experience, including evaluation and benchmarking, is a stated qualification.
Make solutions usable and supported
The role releases validated runbooks and prompt libraries, leads talks and training, and works with broader BXTI to move proofs of concept into supported production deployments while managing concurrent projects.

Likely challenges to think through

Shared demand versus local workflow fit
Likely preparation challenge: a quick fix for one team may not justify a broader platform build. Prepare to explain how you would observe a workflow, test demand with other teams, and choose an appropriately scoped solution before committing engineering effort.
Prototype speed versus production discipline
Likely preparation challenge: an AI proof of concept may demonstrate value before it is reliable, secure, evaluated, or maintainable under real use. Prepare an example of how you would set evaluation criteria and plan the engineering and support work needed for deployment.
Delivery versus adoption across concurrent work
Likely preparation challenge: building several solutions can compete with documentation, training, and production handoff time. Prepare to describe how you would sequence work, communicate scope changes, and tailor guidance for business users and technical teams.

Practice your answers

Technical practice

How would you use Python and SQL to support an AI workflow backed by a data warehouse?

What to coverSuggested answer focus: Explain how you would retrieve, validate, and transform the data needed by the workflow. Discuss query efficiency, data quality checks, and where Python rather than SQL should handle the logic.

How would you troubleshoot an AI-backed service that becomes unreliable under real user load?

What to coverSuggested answer focus: Describe how you would isolate failures, measure their impact, and make a safe fix. Address the posting’s expectations for scalability, reliability, security, and maintained production software.

How would you move a code change through version control, CI/CD, containerization, and cloud deployment?

What to coverSuggested answer focus: Walk through a practical path from review and automated checks to deployment and rollback. Explain what Git, Docker, and infrastructure tooling such as Terraform contribute, without assuming a particular Blackstone setup.

How would you design a retrieval pipeline for a business workflow?

What to coverSuggested answer focus: Cover source selection, retrieval quality, and how users could verify that an answer is grounded in the retrieved material. Discuss trade-offs between coverage, relevance, freshness, and access controls.

When would an agentic workflow be appropriate, and how would you constrain it?

What to coverSuggested answer focus: Explain what a multi-step workflow gains from an agent compared with a simpler approach. Discuss boundaries on actions, error handling, and how you would check whether the added complexity is worthwhile.

How would you choose between prompt engineering and fine-tuning for an LLM use case?

What to coverSuggested answer focus: Compare the evidence needed to justify each approach, including task performance, maintenance effort, and available examples. Describe how you would test a simpler prompt-based approach before committing to a more involved change.

How would you evaluate an applied AI solution before releasing it to users?

What to coverSuggested answer focus: Propose representative workflow tests and benchmarks, including failure cases and criteria tied to the intended task. Explain how you would compare iterations and use feedback after release to find regressions.

How would you decide whether workflow friction calls for a self-service improvement or a larger platform initiative?

What to coverSuggested answer focus: Describe how you would validate demand across teams before building. Compare user impact, repeatability, implementation effort, and ongoing support needs when prioritizing the pipeline.

What would make an AI runbook or prompt library ready for business self-service?

What to coverSuggested answer focus: Explain how you would validate its instructions against the intended workflow and document appropriate use and limitations. Discuss how users could provide feedback and how the material would stay current.

What must change for an AI proof of concept to become a supported production deployment?

What to coverSuggested answer focus: Identify gaps a prototype may leave in testing, operational ownership, documentation, and support under real user load. Explain how you would work with the broader technology organization to close those gaps.

Behavioural practice

Tell me about a time you learned a non-technical team’s workflow before proposing a solution.

What to coverSuggested answer focus: Use a real situation, the steps you took to observe the process and elicit requirements, and the result. Explain what you learned that changed the proposed solution or its priority.

Describe a time you managed competing projects or a change in scope.

What to coverSuggested answer focus: Give the situation, how you weighed priorities and communicated trade-offs, and the result. Include when and how you raised an issue that affected delivery.

Tell me about a time you had to become productive with an unfamiliar technology.

What to coverSuggested answer focus: State the situation, the independent steps you took to learn and verify your approach, and the outcome. Distinguish what you established through testing from what you needed others to clarify.

Describe a time you worked with other technology teams to put a prototype into supported use.

What to coverSuggested answer focus: Explain the situation, your specific contribution to the handoff or production work, and the result. Address how responsibilities and outstanding risks were communicated.

Tell me about a time you explained or taught a technical solution to different audiences.

What to coverSuggested answer focus: Describe the audience and situation, how you adapted a talk, training, or documentation for their needs, and the result. Include how you checked whether people could use the guidance.

Sharpen your knowledge

Topics to prepare

Business workflow discovery
The posting calls for shadowing how teams raise, service, and report on capital, then validating demand before building. As preparation, sketch a hypothetical workflow, mark points of friction, and list questions that would confirm whether the same need exists across teams.
Choosing an LLM approach
The role involves prompt engineering, retrieval, agentic workflows, and evaluation; fine-tuning is listed among relevant experience. As preparation, compare when each approach might fit one hypothetical business request, including what evidence would change your choice.
Evaluation and benchmarking
The team builds evaluation tooling and releases validated AI resources to users. As preparation, define a small test set for a hypothetical retrieval workflow, with checks for answer quality, unsupported claims, and cases that should be escalated.
Production readiness
The posting distinguishes systems maintained under real user load from prototypes and emphasizes scalability, reliability, and security. As preparation, review a hypothetical proof of concept and identify the changes and checks needed before a supported deployment.
Scope and adoption
The role prioritizes self-service improvements alongside larger initiatives, manages concurrent projects, and delivers training. As preparation, rank three hypothetical requests by demand and effort, state what is out of scope, and plan how to explain a scope change.

Skills to revise

Python and SQL data work
The posting requires fluency in both languages for data and data-warehouse work. Practice querying a synthetic warehouse-style dataset with SQL, processing the results in Python, and checking for missing or duplicated records.
Retrieval and agentic workflow implementation
The role builds and extends retrieval pipelines and agentic workflows. As a practice exercise, implement a small retrieval-backed workflow using sample documents, then add a clear handoff or stop condition for requests it cannot answer.
AI test tooling
Evaluation tooling and model benchmarking are named in the posting. Practice writing a repeatable evaluation script for the sample workflow, recording results by test case and comparing two prompt or retrieval variants.
Deployment workflow
The posting names Git, CI/CD, Docker, and Terraform as examples of deployment practices. Practice taking a small Python service through version-controlled changes, automated checks, containerization, and a reviewed deployment plan; no particular cloud provider is specified.
Stakeholder communication and documentation
The role elicits requirements from non-technical colleagues and creates runbooks, prompt libraries, and training for different audiences. Practice a short discovery interview, then turn the notes into a scoped request, a user-facing runbook, and a concise technical handoff.

Your preparation checklist

Map a business workflow
The role involves shadowing how teams raise, service, and report on capital. Prepare a walkthrough of a workflow you know, or a clearly hypothetical one, showing users, friction points, and questions you would ask before proposing AI.
Prioritize possible improvements
The posting calls for validating demand before building. Draft a short decision table comparing a self-service quick win with a larger platform initiative, including who would validate each need and what would change your priority.
Walk through an applied AI project
Choose a project you can discuss, if available, involving retrieval, agents, prompting, or model evaluation. Be ready to explain its design, evaluation results, limitations, and what you would change; label any hypothetical example as such.
Practice a production design exercise
Sketch a Python-and-SQL service that uses warehouse data under real user load. Explain reliability, security, testing, monitoring, and how you would move a proof of concept into a supported deployment; these are stated responsibilities, not a known interview assignment.
Assemble evidence of engineering delivery
Prepare concise examples, where available, of Git-based change control, CI/CD, containerization, cloud deployment, and maintaining software after release. Identify your own contribution and the operational outcome without implying experience you do not have.
Prepare a stakeholder communication example
The role requires scoping, managing concurrent priorities, documenting for different audiences, and delivering training. Outline one relevant example or practice scenario showing a requirements discussion, scope change, issue escalation, and a brief user-facing runbook.
Write focused questions for Blackstone
Ask how BXTI validates demand across BXMA, Private Equity, Real Estate, and Private Wealth; how it evaluates AI solutions before release; and what support is expected after a proof of concept reaches production.

Availability last checked: Nov 4, 2026. Status can change after a check; confirm on the employer’s website. Report an issue · Our checking policy

KEEP EXPLORING

Similar AI jobs

Related skills and specializations in United States of America. Matched to this role, not your profile.

Explore more jobs

Finding similar opportunities…