We create decision infrastructure for the age of strategic surprise. Our Agentic Systems Intelligence (ASI) runs millions of high‑fidelity simulations to surface hidden escalation pathways, giving national‑security, finance, and insurance leaders time to act when minutes matter. Fresh off our AI+ Expo debut, backed by the strongest mission-driven VCs, and featured by Business Insider and NDTV, we’re scaling fast to meet growing demand.

Your Mission

Be the technical and product spearhead of the European engineering hub. Our ASI-powered North Star platform is live - your role is to harden, scale, and expand it.

You will:

Own architecture and delivery alongside our full‑time designer and Chief Scientific Officer (CSO)
By end of year: ship a pilot-ready, secure agentic AI platform that meets Fortune 100/Gov SLOs (latency, reliability, auditability) and passes the pilot security review
Reduce inference unit cost and time-to-value for pilot use cases by ≥30% via architecture, evals, and tooling
Hire, mentor, and inspire the initial engineering cohort

What You’ll Do:

Backend, Compliance, MLOps

Architecture: End-to-end system design for agentic workflows (planning, tool-use, memory, retrieval), multi-tenant isolation, and observability. Author/approve ADRs.
Hands-on delivery: Build core services in Python, APIs, orchestration, data pipelines, eval harnesses; institute CI/CD, IaC, and incident response.
Safety & compliance by design: Prompt/finetune attack resistance, PII handling, audit logs, RBAC/ABAC, data-residency controls, SBOM, supply-chain checks; chart SOC2/ISO27001 paths; support FedRAMP-like/ENS-like customer asks.
MLOps & evals: Offline/online evals (quality, risk, cost), dataset/version governance, model routing, canary & rollback; telemetry for agents (traces, tool calls, failure modes).
Enterprise-grade delivery: SLOs/SLIs, error budgets, on-prem/isolated deployment patterns (including air-gapped variants where needed).

Frontend Development - You are hands-on on the portal/dashboard and set the FE architecture; you will not be the primary research scientist (that’s the CSO’s org)

Design partnership: Co-own the portal IA and dashboard UX with the Product Designer; convert Figma specs into runnable components and document trade-offs in ADRs/“DDRs” (Design Decision Records).
Front-end architecture: Choose and own FE stack (e.g., React/Next.js + TypeScript + Tailwind + Radix/shadcn), component theming via design tokens, SSR/CSR boundaries, BFF (Backend-for-Frontend) layer, real-time data (WebSocket/SSE).
Design system: Stand up a reusable component library + Storybook, visual regression tests, a11y & i18n built-in; enforce versioning.
Agent observability UI: Dashboards for agent traces, tool calls, latency/cost, safety events; drill-downs to reproduce sessions.
Enterprise UX: RBAC/ABAC in UI, audit trails, policy editors, data-residency indicators, customer theming/whitelabel, on-prem/air-gapped safe patterns (no CDN, strict CSP).
Perf & quality: Set FE performance budgets (LCP/INP/CLS), a11y WCAG 2.2 AA, i18n (ICU messages), e2e tests (Playwright/Cypress).

Team Building

Team building: Hire/level the 1-3 additional engineers; mentor strong ICs; set lightweight process.

Tech you’ve likely touched:

LLM/agent orchestration, retrieval, tool calling; vector/RDBMS hybrids; streaming/eventing
Cloud + Kubernetes; IaC (Terraform); secrets/KMS; OTel tracing; staging/prod isolation
Python strongly preferred; Go/Rust a plus. Solid systems and networking instincts
Frontend (portal/dashboard): Next.js (React+TS, SSR/ISR); design system + tokens (Tailwind, Radix/shadcn) with Storybook; data-dense tables & charts; schema-validated forms; i18n + WCAG 2.2 AA; BFF for authz/caching/pagination; real-time traces (WS/SSE); strict CSP, self-hosted assets, feature flags; e2e/unit/visual tests; perf budgets (LCP≤2.5s, INP≤200ms); OTel Web telemetry; offline/zero-egress for on-prem; Workers/WASM for heavy client work.

Tech Stack You’ll Shape

Backend: Python 3.x, FLASK, SQLAlchemy, Celery; APIs with RBAC/ABAC for secure multi-tenant control.
Frontend (Portal/Dashboard): Next.js (React+TS), Storybook, schema-validated forms; real-time agent traces (WebSockets/SSE); accessibility (WCAG 2.2 AA), i18n, strict CSP; safe on-prem patterns.
AI/ML: PyTorch + Hugging Face; RAG pipelines with pgvector/Postgres; agent orchestration (tool use, memory, routing, telemetry).
Infra & DevOps: AWS multi-region (EC2, EKS, S3, IAM) with Terraform/CDK; GitHub Actions; secrets/KMS; on-prem/air-gapped deployment options.
Data & Observability: PostgreSQL + Redis, object storage, Kafka/Pulsar; OpenTelemetry end-to-end tracing; Prometheus/Grafana dashboards; SLO/SLI error budgets.

Must-haves

8–12+ years engineering; 3+ years leading architecture for distributed systems or ML platforms
Comfortable with a fast release cadence and evolving priorities, aligning engineering focus to mission impact.
Proven delivery of secure, customer-facing systems with tight SLOs
Evidence of shipping with researchers/ML teams (you’ve turned novel methods into products)
Excellent writing: crisp ADRs, design docs, runbooks
Shipped data-dense enterprise dashboards with a designer partner
Evidence of design-to-code rigor: tokens, Storybook, visual diff testing
Accessibility & internationalization in production (landed audits)

Nice-to-haves

Prior work with regulated buyers (Gov, defense, F100), data-residency constraints, or on-prem
Security engineering background (threat modeling, secrets lifecycle, SBOM, signing)
Clearance exposure (not required)

Success metrics by Dec 31, 2025

Pilot reliability ≥99.5% success across critical paths; p95 latency targets met
E2E eval suite adopted CI-blocking; safety regression rate goes down quarter-over-quarter
Compute cost per successful task go down ≥30% vs baseline
1-3 engineers hired; on-call and incident playbooks live
LCP ≤ 2.5s, INP ≤ 200ms on pilot hardware; a11y score ≥ 95; visual regressions = 0 at release; task-success rate ≥ 90% for key personas (e.g., Security Officer, Analyst).

How We Work

Triad leadership model: Designer, CSO/CEO, and you form the core decision pod.
Fast feedback: weekly customer demos and design‑research sessions.
Evidence‑driven engineering: trunk‑based workflow, automated tests, DORA metrics.
Remote‑first, Europe‑centric: async by default, with focused overlap for pairing and roadmap reviews.

Why Anadyr

Mission with consequence: your code helps leaders avert conflict and manage systemic risk.
Green‑field autonomy: no legacy baggage—set the patterns that will scale for years.
Hybrid builder‑researcher role: stay close to bleeding‑edge AI/ML while delivering enterprise‑grade reliability.
Backed by deep‑tech investors: long‑term horizon, patient capital, and a network of domain experts.

Next Steps

Please complete the Google Form as your application. Please make sure to answer all questions and upload all requested materials.

Application link: https://forms.gle/2BtAXBGj6SrJUurb8

Expose what others can’t see. Join us and build the platform that lets decision‑makers look beyond the horizon.