Task Designer, Knowledge Work

Reinforced Learning; Town of Tonawanda (NY); On-site; USD 70,000 - 100,000; Full time

Job summary

Reinforced Learning is seeking talent to perform financial analysis, synthesize research, and plan operational tasks. You will design multi-step tasks using documents, data, and tools, and define what a defensible answer looks like.

You will contribute by grounding tasks in real policies and data, co-author rubrics, and ensure a clean separation between training content and evaluation benchmarks. Strong writing and analytical rigor are essential.

Qualifications

• Background in a knowledge-based field: finance, consulting, research, operations, or analytics.
• Ability to write clearly and structure ambiguous problems.
• Proficiency with spreadsheets, SQL, or scripting to build and verify fixtures.
• Commitment to tracing claims to sources.

Responsibilities

• Financial analysis and research synthesis to support decisions.
• Operations planning based on documented data and policies.
• Design tasks requiring multi-step reasoning across documents, data, and tools.
• Create rubrics and evaluation criteria with domain experts.
• Maintain clear boundaries between task design and evaluation data.

Skills

Knowledge professional

Clear writing

Spreadsheets

SQL

Scripting

Evidence tracing

Tools

Spreadsheets

SQL

Scripting

Job description

The role

Financial analysis, research synthesis, operations planning, customer resolution. These are the tasks where models answer well and work poorly. You'll design tasks that require multi-step reasoning over documents, data and tools, and define what a defensible answer looks like.

What you'll own

• Tasks grounded in realistic documents, spreadsheets, records and policies.


• Rubrics that separate a correct answer from a confident one, co-authored with domain experts.


• Document collections and data fixtures with planted conflicts, gaps and distractors.


• Difficulty calibration against expert completion rates and model baselines.


• A clean line between training tasks and held-out evaluation tasks.

What you'll bring

• A background in a knowledge profession: finance, consulting, research, operations, analytics.


• You write clearly and structure ambiguous problems well.


• Spreadsheets, SQL or scripting to build and check fixtures.


• Respect for evidence. You want claims traced to sources.

Helpful, not required

• Assessment, case study or examination design.


• Familiarity with how language models are evaluated.