AI Solutions Engineer

• Own technical onboarding for new customers: data ingestion, indexing, metadata mapping (jurisdiction, authority, recency), and validation.
• Configure and adapt retrieval + reranking pipelines to customer needs (practice area focus, doc structure, internal taxonomies, "what good looks like").
• Implement customer-specific workflows: templates, filters, jurisdiction defaults, citation behavior, permission-aware retrieval, and custom result layouts.

LLM workflows that are production-safe

• Tailor prompting / context engineering to customer requirements (traceability, citation style, explanation depth, fallback behavior).
• Implement safeguards: provenance, source grounding, "no-citation * no-claim" behaviors, and "confidence/uncertainty" patterns aligned with legal risk.

Evaluation & iteration in the field

• Build lightweight customer-specific eval sets (gold questions, acceptance criteria, "must-not-fail" cases).
• Run fast error analyses and ship fixes: query understanding tweaks, reranker tuning, chunking strategy, dedupe suppression, caching, and retrieval routing.

Performance, cost, reliability

• Keep latency and costs under control with caching, batching, early exit, and sensible fallbacks.
• Monitor quality + usage signals; turn customer feedback into concrete improvements and measurable acceptance checks.

Collaboration & knowledge transfer

• Work closely with Customer Success + legal experts to translate pain points into system changes.
• Document integrations and "deployment recipes" so solutions become reusable product capabilities over time.

Do you enjoy taking a strong AI product and making it work beautifully inside a customer's real-world environment; messy data, unique workflows, strict permissions, and high expectations? Are you hands-on, pragmatic, and happiest when you can ship an improvement that a specific legal team immediately feels?

You're comfortable being the "technical bridge" between customers, legal experts, and the core product team: you diagnose issues, propose solutions, implement them, and leave behind clean playbooks so the next deployment gets easier., * Strong hands-on experience building or adapting search/retrieval systems in production (hybrid retrieval, reranking, query understanding, indexing).

• Proven experience taking LLM workflows from prototype to reliable production use.
• Proficiency in TypeScript/Node.js (our core stack).
• Experience with one or more of: Azure AI Search, pgvector/PostgreSQL, OpenSearch/Elasticsearch (or similar).
• Practical engineering instincts: debugging, performance tuning, careful handling of edge cases, and clear operational thinking.
• Strong communication skills and comfort working directly with customers (technical deep-dives, explaining trade-offs, writing playbooks).
• Proficiency in English; full-time availability.
• Hybrid presence: on-site in Zurich at least two days per week., * German proficiency (many sources and customer interactions are German-speaking).
• Experience integrating customer data sources / document pipelines (connectors, ETL, access controls).
• Experience with pragmatic eval pipelines (human-in-the-loop labeling, inter-annotator agreement, lightweight dashboards).
• Familiarity with sparse + dense retrieval methods (BM25 variants).
• Experience operating services (Docker is a plus).
• Familiarity with our stack: Azure / NestJS / Next.js.
• Knowledge of Swiss / German / US legal systems is a plus.

• Customer-visible impact: your work directly determines whether customers trust the product in daily legal workflows.
• Autonomy & ownership: you'll own deployments end-to-end and shape repeatable "solution patterns."
• Fast learning loop: see real-world failure modes early; help steer product priorities with evidence.
• Compensation: CHF 7'000-11'000 per month + ESOP, depending on experience and skills.

Omnilex is a young dynamic AI legal tech startup with its roots at ETH Zurich. Our interdisciplinary team (12+ people) empowers legal professionals by leveraging AI for legal research; combining external data, customer-internal data, and our own AI-first legal commentaries.