LLM Integration Engineer (Remote - Contract)
💡 Søknadstips: Når du klikker på «Søk gratis på Braintrust», blir du videresendt til den offisielle siden til Braintrust. Dette er 100 % gratis for deg og bidrar til å støtte plattformen vår gjennom henvisningsbonuser.
⚠️ Oversettelsesmerknad: Denne stillingsinformasjonen er oversatt med AI. Ved uklarheter eller feil, bruk den engelske originalen som fasit.
Role Overview
IMP Intelligence is a private compensation data company. We maintain a proprietary, structured database of compensation benchmarking data across the US, UK, and Europe. Our platform is used by well-known financial services firms, so this work needs to be accurate, reliable, and production-ready.
We're building an AI-powered experience that lets clients query our compensation database in natural language and receive accurate, formatted benchmarking outputs (e.g., percentile ranges). Our client-facing front end is built in Bubble and will remain in Bubble, but the backend must be custom-built outside Bubble (SQL-backed) and connected via API. Authentication/gated access already exists within our portal.
This is a remote contract role. We're hiring one person to own delivery end-to-end with minimal hand-holding.
Responsibilities:
- Build a backend service that connects our structured compensation data to an LLM via API (e.g., GPT-4, Claude, or another appropriate model) to support natural language queries
- Implement a reliable retrieval approach (e.g., NL → structured query) that accurately pulls the correct data from our private database
- Build a custom backend + SQL data layer (outside Bubble) and expose an API that our Bubble front end can connect to (endpoints for query input and formatted output)
- Return clean, client-ready benchmarking outputs (on-screen; PDF export is a plus)
- Implement safe fallback behavior when direct data is missing (e.g., "insufficient data" or clearly labeled proxy ranges) with guardrails to avoid hallucinations
- Ensure the solution is production-ready: stable performance, clear error handling, and maintainable code
- Communicate progress and tradeoffs clearly to a non-technical product owner (plain English explanations)
Requirements (Non-negotiable)
- Demonstrable experience building LLM-powered applications connected to structured databases (e.g., LangChain, LlamaIndex, RAG patterns, NL → SQL workflows, etc.)
- You have shipped a relevant product and can prove it (live product, repo, or demo + walkthrough)
- Experience building backend APIs (REST) that integrate with external front ends (Bubble-compatible API experience is a strong plus)
- Ability to work independently and own delivery (requirements → build → iterate → ship)
- Strong written and verbal communication in English
Nice to have
- PDF report generation / templating experience
- Experience building with private/proprietary datasets and implementing guardrails to avoid hallucinations
Få personlige jobbvarsler