Top Benefits
About the role
About us Haste AI is a strategically positioned vertical AI company deploying and developing CoEngen, our proprietary AI platform purpose-built for the mining industry.
At Haste AI, you’ll work at the forefront of software development and specialized AI, creating intelligent systems with highly proprietary and sensitive data, that drive efficiency, safety, and sustainability in one of the world’s most critical sectors. CoEngen is a cutting-edge proprietary software that is revolutionizing how the mining industry harnesses data, optimize processes, and make smarter decisions in complex environments.
The thing that makes CoEngen work is a hard rule: every number is computed by deterministic code, and the AI only writes prose around numbers it is handed. The two never cross. A wrong figure in a signed technical report is not an option, so correctness is not a nice-to-have here, it is the product.
The role This is a founding engineering hire. We are still small, and you would help set the technical direction of the AI platform rather than just build tickets against someone else's plan. You would own big pieces end to end: the agent that drafts reports, the guardrails that keep it honest, and the infrastructure that lets us run all of it where our customers need it to run. Our clients handle sensitive, non-public geological and financial data, so a lot of the intersting work right now is making sure that data never leaves where it is supposed to. You will work directly with the founder and with the people actually using CoEngen, and your equity will reflect that you got here early.
What you'll work on The report-drafting agent: tool design, grounding, and the number-integrity guardrails that stop the model from ever inventing or altering a figure. Running LLMs in-region and inside customer environments: in-region managed inference (Bedrock, Azure OpenAI, Vertex) as the main path, with self-hosted inference where a customer needs it. A de-identification and scrubbing layer so only scrubbed data is ever put in front of a model, plus the data-residency and zero-retention controls around it. The deterministic compute engine and ingestion pipeline that produce every number the reports rely on (TypeScript, Bun, Hono, Zod, Drizzle, Postgres). Helping us reach and hold SOC 2, and answer the hard security questions that enterprise and regulated customers ask.
What we're looking for You have shipped real LLM-backed features to production. Agents and tool use, retrieval, grounding, evals. Not just wrapping a chat endpoint. You have deployed models in regulated or restricted settings and understand data-residency and retention constraints: in-region managed inference (Bedrock, Azure OpenAI, Vertex), VPC isolation, and knowing what can and cannot leave a customer boundry. Strong typed backend engineering. TypeScript or Node in production and real comfort with Postgres. Our stack is Bun, Hono, Zod, Drizzle, and Turborepo, which is a plus, but we care more that you write clear, well-tested code. A real instinct for data privacy: PII and identifier handling, de-identification, and least privilege. You are at home in correctness-critical systems. You reach for tests, determinism, and verification because you have been burned before by systems that were probably fine.
Bonus points Self-hosted or on-prem inference experience (vLLM, TGI, Ollama, llama.cpp), including GPU basics, quantization, and cost and latency tuning. We do not need this on day one, but it is a real plus. You have worked in a regulated or high-stakes domain (mining, geoscience, finance, health, defense). SOC 2, ISO 27001, or similar compliance work. Willingness to dip into the frontend (Next.js, React) when a feature needs it end to end.
How we work Small team, high trust, low process. You own things end to end and talk directly to the people using CoEngen. We care about getting the numbers right, shipping, and not wasting each other's time.
Details Type: Full-time, remote Equity: Meaningful founding-level equity, negotiable
About Haste Ai
HASTE Ai builds CoEngen, a system that turns raw metallurgical and comminution testwork into securities-grade technical reports (NI 43-101, ISO 17025). Mining companies and labs spend weeks hand-writing these reports and checking every number by hand. We do it in a fraction of the time without giving up an ounce of accuracy.
Similar Jobs
Top Benefits
About the role
About us Haste AI is a strategically positioned vertical AI company deploying and developing CoEngen, our proprietary AI platform purpose-built for the mining industry.
At Haste AI, you’ll work at the forefront of software development and specialized AI, creating intelligent systems with highly proprietary and sensitive data, that drive efficiency, safety, and sustainability in one of the world’s most critical sectors. CoEngen is a cutting-edge proprietary software that is revolutionizing how the mining industry harnesses data, optimize processes, and make smarter decisions in complex environments.
The thing that makes CoEngen work is a hard rule: every number is computed by deterministic code, and the AI only writes prose around numbers it is handed. The two never cross. A wrong figure in a signed technical report is not an option, so correctness is not a nice-to-have here, it is the product.
The role This is a founding engineering hire. We are still small, and you would help set the technical direction of the AI platform rather than just build tickets against someone else's plan. You would own big pieces end to end: the agent that drafts reports, the guardrails that keep it honest, and the infrastructure that lets us run all of it where our customers need it to run. Our clients handle sensitive, non-public geological and financial data, so a lot of the intersting work right now is making sure that data never leaves where it is supposed to. You will work directly with the founder and with the people actually using CoEngen, and your equity will reflect that you got here early.
What you'll work on The report-drafting agent: tool design, grounding, and the number-integrity guardrails that stop the model from ever inventing or altering a figure. Running LLMs in-region and inside customer environments: in-region managed inference (Bedrock, Azure OpenAI, Vertex) as the main path, with self-hosted inference where a customer needs it. A de-identification and scrubbing layer so only scrubbed data is ever put in front of a model, plus the data-residency and zero-retention controls around it. The deterministic compute engine and ingestion pipeline that produce every number the reports rely on (TypeScript, Bun, Hono, Zod, Drizzle, Postgres). Helping us reach and hold SOC 2, and answer the hard security questions that enterprise and regulated customers ask.
What we're looking for You have shipped real LLM-backed features to production. Agents and tool use, retrieval, grounding, evals. Not just wrapping a chat endpoint. You have deployed models in regulated or restricted settings and understand data-residency and retention constraints: in-region managed inference (Bedrock, Azure OpenAI, Vertex), VPC isolation, and knowing what can and cannot leave a customer boundry. Strong typed backend engineering. TypeScript or Node in production and real comfort with Postgres. Our stack is Bun, Hono, Zod, Drizzle, and Turborepo, which is a plus, but we care more that you write clear, well-tested code. A real instinct for data privacy: PII and identifier handling, de-identification, and least privilege. You are at home in correctness-critical systems. You reach for tests, determinism, and verification because you have been burned before by systems that were probably fine.
Bonus points Self-hosted or on-prem inference experience (vLLM, TGI, Ollama, llama.cpp), including GPU basics, quantization, and cost and latency tuning. We do not need this on day one, but it is a real plus. You have worked in a regulated or high-stakes domain (mining, geoscience, finance, health, defense). SOC 2, ISO 27001, or similar compliance work. Willingness to dip into the frontend (Next.js, React) when a feature needs it end to end.
How we work Small team, high trust, low process. You own things end to end and talk directly to the people using CoEngen. We care about getting the numbers right, shipping, and not wasting each other's time.
Details Type: Full-time, remote Equity: Meaningful founding-level equity, negotiable
About Haste Ai
HASTE Ai builds CoEngen, a system that turns raw metallurgical and comminution testwork into securities-grade technical reports (NI 43-101, ISO 17025). Mining companies and labs spend weeks hand-writing these reports and checking every number by hand. We do it in a fraction of the time without giving up an ounce of accuracy.