Prompt/ML Engineer

Doczen; Remote; GBP 80,000 - 100,000; Full time

Job summary

A vertical-AI workflow platform is seeking a Prompt Engineer to join their innovative team. This hands-on role involves complex prompt engineering to enhance AI-driven workflows in the commodities sector. You will design workflows to transform unstructured emails into structured, actionable data, collaborate with engineers to integrate solutions, and continuously improve model behavior. Candidates should possess strong Python and AI/ML skills and embrace a builder mindset to thrive in this dynamic remote-first environment.

Benefits

Competitive compensation

Flexible time off

Learning stipend for professional growth

Qualifications

• Proven experience shipping complex prompt-engineered systems in production.
• Strong Python skills, comfortable reading/writing code for integrating prompts.
• Solid AI/ML background focused on NLP and information extraction.

Responsibilities

• Own complex prompt engineering end-to-end with structured outputs.
• Design and iterate workflows to extract and validate data from unstructured sources.
• Collaborate with engineers to integrate prompts into services and frontends.

Skills

Complex prompt engineering

Python

AI/ML knowledge

Curiosity about model behavior

Tools

Python services

React/TypeScript

Job description

Join a fast, founder-led team building an inbox-native operating fabric for the commodities sector—where AI transforms everyday communications into reliable, scalable workflows. This is a hands‑on role for a Prompt Engineer with proven experience in complex prompting, Python, and applied AI/ML, shaping how LLMs orchestrate mission‑critical commodity operations with room for outsized impact and growth.

Why Doczen

Doczen is a vertical‑AI workflow platform that meets teams where work already happens—inside email and spreadsheets—so operations gain structure and automation without heavy change management. The long‑term vision is “EDI for everyone”: unifying communications, data, and decisions through agentic workflows so teams and partners always have the right information at hand.

The Opportunity

In physical commodities, a single voyage can involve dozens of counterparties, 70+ documents, and 400–500 emails—most of it unstructured and time‑sensitive. Doczen bridges the gap between messy email‑based operations and rigid EDI with AI that reads, routes, and verifies information while letting users keep their familiar tools, and your prompts, evaluations, and workflows will sit at the core of that reliability.

What You’ll Do

• Own complex prompt engineering end-to-end: multi‑step agents, retrieval strategies, tool use, structured outputs, and guardrails for production AI features.
• Design and iterate prompts and workflows that reliably extract, normalize, and validate data from unstructured emails and documents across high‑stakes commodity operations.
• Build and maintain robust evaluation harnesses (golden datasets, regression tests, quality metrics, telemetry) to measure and improve model behavior over time.
• Collaborate with engineers to integrate prompts into Python services, React/TypeScript frontends, and workflow orchestration with idempotency, retries, backoff, and dead‑letter handling.
• Work with customers and internal stakeholders to understand edge cases and failure modes, translating them into better prompts, data contracts, and evaluation strategies.

What Success Looks Like (30–60 Days)

• 30 days: Shipping prompt and workflow changes that improve extraction accuracy and stability on real customer traffic, backed by clear evaluation signals.
• 60 days: Actively shaping the backlog for AI features, leading experiments, and driving measurable gains in reliability, speed, and user trust.

Must-Have Experience

• Proven experience shipping complex prompt‑engineered systems in production (not just prototypes), including RAG, tool‑calling, and structured JSON output contracts.
• Strong Python skills, with comfort reading/writing code to wire prompts into services, pipelines, and CI, plus familiarity collaborating in a modern engineering stack.
• Solid AI/ML background (e.g., NLP, information extraction, applied ML) and a deliberate transition into prompt engineering as a primary craft.
• A passionate, builder mindset: high ownership, speed, and impact, with curiosity to deeply understand model behavior and real‑world operational constraints.

Nice to Have

• Experience in commodities, supply chain, or other intricate B2B domains where data is messy and workflows are high‑stakes.
• Experience with performance tuning and telemetry for AI‑powered features and distributed services.

How We Work

Remote‑first with a fast, collaborative cadence; as we scale, we expect to centralize more around customer proximity while keeping strong async habits. High ownership, high trust, and tight feedback loops define how we operate, and above all we work hard, joyfully, and have fun while building.

Compensation & Benefits

We believe in pay transparency—based on seniority we maintain salary bands of I, II, and III (hence the posted salary range), and equity ranges are shared early in the process. Expect competitive compensation with equity, flexible time off, and a learning stipend to support continuous growth in AI, ML, and prompt engineering craft.

Interview Flow

A clear, predictable process: intro screen → half‑day deep‑dive → same‑day decision, respecting your time and giving you space to see how we work in practice.

About Doczen

Doczen structures unstructured communications and attachments into auditable, automated workflows—right from the inbox—so operations teams move faster with less friction. The trajectory is “EDI for everyone,” with agentic automation and deep integrations that support the sector’s most critical decisions while keeping the user experience simple and familiar.

Apply if this sounds like you

Builders who love shipping, own outcomes, and thrive on clarity of impact will find outsized scope and leverage here—especially if you are genuinely passionate about LLMs and complex prompt engineering. If you’re founder‑minded, excited by hard AI/operations problems, and ready to figure things out with a fast, dynamic team, this role is for you.