AI Evaluation, Test and Assurance Scientist

About Precision Federal

Precision Federal is a Christian-owned small business building AI and software systems for federal agencies through the SBIR program. The founder is a Christian, and the company is run on Christian principles: every number we publish is real, every claim we make is verifiable, and the work is done with excellence as service to God. Every proposal we submit is backed by working prototypes built before we ask for a dollar.

How this role works

1. You join the bench. We name you (with a short bio) as a proposed consultant on proposals that match this role. Consent costs you nothing and commits you to nothing until an award funds.
2. A contract is awarded. Federal SBIR awards fund the team named in the proposal. Your engagement begins, at an hourly rate agreed before you were ever named.
3. You work remote, paid from the contract. ~40–80 hrs per funded award — and if no award ever funds, you owe nothing and lose nothing.

What you'll do

• Design metrics, test sets and benchmarks for AI systems, including explainability, robustness and influence or bias measures
• Build reusable evaluation tooling that runs against sensor, text and multimodal models
• Write the assurance case and the test report a DoD T&E organisation can act on
• Point us at the standards (NIST AI RMF, DoD RAI, ISO/IEC 42001) a reviewer will expect

Required qualifications

• Direct experience evaluating machine-learning systems beyond accuracy: robustness, calibration, drift, adversarial and human-influence testing
• Statistics and experimental-design background you can defend
• U.S. citizen or lawful permanent resident (U.S. person)
• Willing to be named on a federal proposal with a one-page resume, and to sign a short letter of commitment

Preferred qualifications

• DoD test and evaluation or operational test experience
• LLM red-teaming and benchmark authorship
• Publications in ML evaluation or trustworthy AI

Compensation & structure

1099 consulting engagement. $60–$100 per hour, confirmed in writing before you are named on any proposal. Hours are funded by federal contract awards — engagements on this bench have ranged from roughly 130 to 710 hours across a period of performance, fully remote. Work and pay begin only if the contract is awarded — no award, no hours, no obligation.

Apply — AI Evaluation, Test and Assurance Scientist

Every application is read personally by the founder. We respond to every serious one.

First name *

Last name *

Email *

Phone

City & state

Years of relevant experience * Select Under 5 5–10 10–15 15–20 20+

Work authorization * Select U.S. citizen U.S. lawful permanent resident Other

Security clearance Select (optional) None Prior (inactive) Active

Military service Select (optional) None Veteran Retired military Currently serving (Guard/Reserve)

LinkedIn or portfolio link

Resume (PDF or Word) *

How did you hear about us?

Five questions

These are the substance of the application. Answer from your own experience — a few sentences each is plenty, and “this never came up” is a real answer we want.

1. How often did an AI model that passed its own benchmark and failed the first week in the field come up in your work, and roughly what did it cost — in hours, in people, or in rework? *

2. What tool or process does your organization use today for AI performance evaluation, benchmarking and assurance, and where does it fall short? *

3. If a better way to handle AI performance evaluation, benchmarking and assurance existed, which office or role would decide whether to adopt it? *

4. Do you have any role in deciding what tools or systems your organization adopts — approving, evaluating, or recommending a purchase? *

5. Do you own, or hold signing authority for, a company or consultancy of your own? *

May we name you, by name and title, as a proposed consultant on our team in federal proposals, and quote your answers above? * Please choose… Yes No

Submit application

Questions before applying? contact@precisionfederal.com · See all open roles