AI Engineer Poland II

Sqope Intelligence, an enhanced due diligence services company established in 2010 with offices in Luxembourg, Geneva, and London, has opened an Israeli subsidiary (Sqope AI) and is looking to invest in the development of AI as a means of enhancing and augmenting the capabilities of its intelligence and research division.

This subsidiary is looking for an AI engineer to work directly under the CTO to help build and strategize the AI tool.

This AI Engineer will live at the intersection of LLMs, retrieval, and API-first microservices. You’ll own end-to-end features: from prompt pipelines and long-context strategies to self-hosted LLMs and production APIs. The position is remote for now and will eventually move to a hybrid model.

The ideal candidate will have the following skills:

• LLM stack: LangChain/LangGraph, OpenAI/Anthropic, vLLM/Ollama/TGI, embeddings (Text-Embedding, SGPT/BGE), rerankers (ColBERT/Cross-Encoders)

• Services: FastAPI, Node/TypeScript, WebSocket/Server-Sent Events for streaming.

• Data & search: PostgreSQL/PGVector, Redis, S3/MinIO, optional graph DBs (Neo4j/Arango) or in-house KG

• Infra: Docker, Kubernetes/EKS or ECS Fargate, Terraform, CI/CD (GitHub Actions/CircleCI), metrics/logging (Prometheus/Grafana/Datadog), OpenTelemetry.

• Design & ship LLM features: Build robust agents, tools, and pipelines using LangChain/LangGraph (our de-facto framework) with a focus on reproducibility, observability, and testability.

The candidate will be responsible for the following:

• Creating strategies to manage the context window: Implement chunking, hierarchical and hybrid retrieval, graphs, query planning, windowed conversation memory, and citation-aware responses.

• Hosting private LLMs at scale: Stand up and optimize self-hosted models (e.g., vLLM/Ollama/TGI) for latency, throughput, and cost; evaluate model choices (open vs. hosted) per use case.

• Creating APIs & microservices: Build clean, versioned FastAPI/Node services; define contracts; add auth, rate limits, and telemetry; package with Docker and deploy

• Handling Evaluation and quality: Set up offline/online evals, golden sets, regression tests, and human-in-the-loop review; track outcome metrics (accuracy, grounding, latency, cost).

• Working Security First: Enforce strict data boundaries and audit trails for tenant-isolated contexts; integrate PII handling, redaction, and policy-driven access controls

• Being part of product strategy and decisions to shape the tool maximizing AI capabilities

• Being able to work both independently and as part of a team within a largely remote setting, managing their own calendar and deadlines

• Having full professional fluency in English