Forward‑Deployed AI Engineer (LLM/ML)

At Silent Eight, we develop our own AI-based products to combat financial crimes that enable things like money laundering, the financing of terrorism, and systemic corruption. We’re a leading RegTech firm working with large international financial institutions such as Standard Chartered Bank and HSBC. Join us and help make the world a safer place!

We solve hard, real‑world problems — from uncovering financial crime, fraud patterns and mule networks, through prioritising thousands of alerts, to crafting defendable case narratives. We work close to users (analysts, investigators, risk/compliance), iterate fast, and deliver in weeks, not quarters. The adversary adapts — this is an intelligence game, not an academic benchmark.

The Role
We’re looking for someone who solves business problems with technology. Less stack worship, more outcomes: fast risk identification, fewer false positives, faster time‑to‑decision, better explainability, and lower cost per case. Finance shows up often, but we think broader — investigations, decisions and client value across industries.

What you’ll do
• Go to the field: talk to users, shadow their workflows, capture the as‑is → goals & constraints.
• Define hypotheses & KPIs (precision/recall, FPR, TAT, coverage, cost/decision) and turn them into experiment plans.
• Design decision flows that mix LLMs, retrieval/RAG, classical ML, and lightweight rules; ensure explainability and auditability.
• Build quick prototypes (notebook → lightweight service/API) and measure their impact on real data.
• Create evaluation sets and scoring rubrics (offline + side‑by‑side + sanity checks + guardrails).
• Present findings & recommendations directly to decision‑makers; propose rollout (pilot → production‑lite → scale).
• Lead innovation processes across the company; test, promote solutions and mentor others with new AI technologies.


Minimum Requirements
• A track record of delivery: 2–3 examples where your AI/ML solution materially improved process KPIs (any industry).
• Problem‑solving & communication: you can break down fuzzy problems and explain risks to non‑technical stakeholders.
• LLMs + ML in practice: RAG, prompting, tool‑calling; classification/ranking/deduplication; fundamentals of evaluation & experimentation.
• Python + SQL sufficient to build a prototype that works and can be maintained.
• Polish & English fluency for user conversations and concise write‑ups.


Nice to Have
• Experience with investigations / trust & safety / fraud / risk / audit or other complex decision processes.
• Graphs/ER: entity resolution, link analysis, pattern‑of‑life.
• Light engineering craft: FastAPI, Docker; the rest (K8s/CI/CD/Terraform) is not required.

Our Tech (lightweight)
We don’t fetishise the stack. Common tools: Python, SQL, notebooks/analysis tools, lightweight APIs (e.g., FastAPI), simple stores (e.g., Postgres), and vector indexes. We choose tools pragmatically — business impact beats heavy infrastructure

Note: we don’t expect mastery of “every” tool. What matters are strong fundamentals, curiosity, and a habit of delivering measurable outcomes.


How We Work
A small, seasoned squad. Weekly goals and regular weekly iterations, a build → measure → learn rhythm. Minimum ceremony, maximum user contact. Decisions are captured briefly — in ADRs, tickets, or short notes.

What We Offer

• B2B contract with Fully-remote role, forever.
• Employee stock option scheme.
• Paid days to balance work and personal life.
• Clear career path, promotion, and development opportunities in a global organization.
• Real influence on decisions in matters that actually count.
• Forward‑deployed autonomy — you stay close to the problem and solve it end‑to‑end. 

Recruitment process:

1. Phone Screen with People Ops (Karol) - 30 mins

2. Technical interview with our CTO (Michael) - 30 mins

3. Home assignment - Leave 5 calendar days to the candidate

4. Hiring Manager Interview with Saurabh - 20 minutes