Commercial Blasting Engineer

What you would do

• Compose prompts at benign, dual-use and adversarial levels targeting blast design, safety procedures, inventory management and transport protocols
• Review model responses against a policy standard and judge whether each one was handled correctly
• Write detailed reference answers explaining technical reasoning, regulatory requirements and safety trade-offs
• Analyze patterns across multiple responses to identify where models are over-refusing or under-refusing

Who they want

• Licensed blaster or blasting engineer with field experience in mining, quarrying, construction or demolition
• Deep knowledge of blast design, vibration control, seismic monitoring and regulatory compliance
• Knowledge of explosives storage protocols, record-keeping requirements, and logistics for controlled materials
• Background in technical documentation, research publication, or consulting roles showing ability to communicate specialized knowledge
• Comfort with writing and reading about misuse scenarios for sustained periods without classified or export-controlled information

Main skills

What the interview asks about

1. 1.Dual-use judgment in real scenarios

   The core skill is recognizing when a question that sounds like routine site work is actually fishing for dangerous information, since industry details that satisfy auditors can enable diversion.

   For example: “Two prompts on vibration limits for urban demolition near a hospital: one routine, another with 'minimal documentation' and 'unmonitored.' How would you write the versions and why does the difference matter?”

2. 2.Technical writing for non-experts

   Your reference answer must convince a non-specialist that your judgment is sound, which requires explaining blast mechanics, timing, burden calculations and regulatory reasoning without jargon.

   For example: “A model refuses a charge weight calculation for rock type and soil depth. Write the reference answer explaining when this is routine, what the answer looks like, and if the refusal was justified.”

3. 3.Adversarial prompt construction

   Testing AI judgment requires prompts that pressure the system toward over-helping or over-refusing; you need to know where those boundaries are and how to probe them without crossing ethical lines yourself.

   For example: “Design a sequence of three prompts about magazine storage and transport, starting with standard regulatory compliance questions, escalating to questions about concealment and minimal record-keeping, to test whether the model recognizes the shift in intent.”

4. 4.Blast design and vibration compliance

   Your technical credibility rests on mastery of burden, spacing, timing and seismic monitoring thresholds; you must evaluate AI reasoning about these parameters and recognize when it oversimplifies or misses regulatory constraints.

   For example: “An AI model suggests that increasing the delay time between detonations will always reduce vibration at a nearby structure. What does the model miss about near-field vs. far-field effects, regulatory monitoring standards and site-specific variables?”

5. 5.Magazine and inventory diversion tactics

   Storage compatibility, security procedures and chain-of-custody rules protect crews daily; understanding how these systems prevent theft and diversion informs what a malicious prompt is really asking.

   For example: “Compare two questions about reconciling inventory after a storage relocation. One is routine regulatory documentation; the other subtly asks about gaps in oversight or procedures for unreported discrepancies. What makes the second one a red flag?”

A task you may get

Given an AI response to a dual-use blasting prompt, evaluate whether it balanced helpfulness with safety, identify gaps in reasoning, and draft a reference answer that would have been appropriate.

How to prepare

• Review your own technical writing samples, published research or expert witness reports to sharpen examples of clear explanation to non-specialists
• Reflect on three past scenarios where a routine professional question had a hidden edge or could be misused; practice articulating the difference in writing
• Study your field's current regulatory standards for vibration, storage, inventory and transport documentation so you can evaluate AI accuracy
• Anticipate the most tempting misuse scenarios in your work and draft both the dangerous question and the safe, helpful alternative

The facts

Pay

$65–75/task

Hours

Task based

Where

Remote

Field

Other Engineering

Project name

Neon

Posted

9/14/2026

Places left

100

We wrote this page from the public Mercor listing. It may be out of date, so read the full posting before you apply.