Data & Eval Operations Program Manager
- Salary
- $250K–$450K
- Moves you to
- United States
- Support
- Visa sponsorship
- Posted
Show job descriptionHide job description
The Role
As the Data & Eval Operations Program Manager for our team, you will structure and implement the processes for data acquisition and eval execution for our foundation models and will develop these into comprehensive programs. You will work closely with our techincal team, including research scientists, research engineers, and software engineers, as well as with external data vendor partners as our key interface. These programs will have a massive impact on the continuous improvement of the models we develop, meaning that this is a critical role to the ultimate success of the organization. You will be expected to stand up and manage these programs autonomously, and tailor them over time to the changing needs of the team.
Key Responsibilities
Data Operations
-
Establish Data Ops Program: Establish a complete end-to-end data operations program from scratch, to include program structure, goals, and management. You will be completely responsible for standing up and owning this program.
-
Understand the Landscape: Analyze and develop an understanding of the constantly evolving landscape of data offerings from new and existing vendors, and assess how this fits the evolving needs of IFM with technical leads.
-
Vendor Management: Work closely with new and existing vendors on a regular basis. You will serve as the key interface between IFM and our data vendors.
-
Budget and Performance Management: Perform budget management and tracking, as well as performance management. You will be responsible for the full data budget and will devise a rigorous process to ensure this spend is highly effective, with all spend tied back to actual data performance.
-
Cross-Functional Collaboration: Work closely with cross-functional teams, including Legal and Procurement, to ensure new data partnerships and acquisitions are completed quickly and in alignment with all organizational procedures.
-
Establish Eval Ops Program: Establish a complete end-to-end eval operations program from scratch, to include program structure, goals, and management. You will be completely responsible for standing up and owning this program.
-
Eval Process: Devise a process for how our team regularly completes, documents, organizes, and reviews our evals.
-
Eval Process Changes: Work with technical leads to understand their evolving eval needs and adjust the eval process to meet these needs.
-
Streamline Eval Service: Streamline the eval process and internal eval service with our engineering team and with individual eval owners to improve how evals are completed.
-
Tie Evals to Data: Develop a sense of where our models might be lagging with respect to key evals and use this to inform strategy for the data operations program.
-
Understand the Landscape: Analyze and develop an understanding of the constantly evolving landscape of benchmarks that may be of interest to IFM.
Eval Operations
Academic Qualifications
Professional Experience - Minimum
-
5+ years of relevant industry experience, with at least 3 years in program management with technical teams, including working directly with researchers and engineers.
-
Budget Management: Budget management experience, managing multi-million-dollar budgets.
-
Vendor Management: Vendor management experience, working across multiple vendors, and experience with contracts and SOWs.
-
Complete Program Ownership: Autonomy in fully managing projects with budget, timeline, and quality components.
-
Technical Fluency: You don’t need to write code or understand nuanced research but must understand relevant technical concepts.
Professional Experience - Preferred
-
Foundation Model Experience: Experience working specifically with LLMs, Generative AI, or ML Infrastructure.
-
Data Operations Experience: Experience working with data vendors for LLM development.
-
LLM/Agentic Benchmarks Understanding: Familiarity with key LLM/agentic benchmarks, including an understanding of how LLMs, aagentic capabilties, and benchmarks are evolving over time.