Senior AI Engineer (up to £140k)

Dex; Greater London; Sur place; GBP 90 000 - 130 000; Plein temps

Résumé du poste

Dex is seeking a Senior AI Engineer to own the core agent harness end to end, delivering streaming graph execution and interruptible tool use. You will build end-to-end autonomous agents that can research companies, digest complex documents, and provide verifiable, cited analyses for high-stakes diligence.

You will craft proprietary infrastructure, multimodal retrieval pipelines, and robust evaluation suites to ensure reliability, with close collaboration to founders and PE experts in a

Qualifications

• 5+ years building and scaling production backend services.
• Proven track record taking LLM features or complex retrieval systems to production.
• Expertise in Python and production search/data infra (OpenSearch, PostgreSQL, FastAPI).
• Eval-driven mindset with benchmarking, verification and error analysis.

Responsabilités

• Design, ship, and scale autonomous agents that research companies and generate verifiable, cited analysis.
• Build proprietary foundation infrastructure, including a streaming-first graph execution framework and an interruptible, resumable code execution environment for tool use.
• Develop multimodal retrieval pipelines combining vision-language models for OCR and deep document search.
• Engineer production eval suites and feedback loops that serve as gate for harness modifications and model updates.
• Collaborate with founders and private equity domain experts to turn complex diligence workflows into reliable programmatic systems.

Connaissances

Backend systems

Production AI experience

Python

OpenSearch

PostgreSQL

FastAPI

Distributed queues

Outils

OpenSearch

PostgreSQL

FastAPI

Message queues

Description du poste

The role

Most AI tools in finance are brittle prompt wrappers over generic models. This Series A startup builds native agentic infrastructure for multi-billion-dollar private equity funds, automating technical sourcing, diligence, and portfolio analysis with fully cited audit trails. Backed by top fintech and venture investors and growing revenue more than ten-fold over the past year, they have achieved clear product-market fit with top-tier buyout and growth equity deal teams across the US and Europe.

You will join as a Senior AI Engineer owning the core agent harness end to end, from execution internals and retrieval pipelines to the evals that prove system reliability. This role is not about stitching together off-the-shelf framework templates or tweaking demos for show. It is about engineering streaming graph execution, interruptible tool calling, and multimodal document search robust enough that investment committees trust it with nine- and ten-figure decisions.

The work

• Design, ship, and scale autonomous agents that research companies, digest complex financial documents, and generate verifiable, cited analysis.
• Build proprietary foundation infrastructure instead of relying on generic wrappers, including a streaming-first graph execution framework and an interruptible, resumable code execution environment for tool use.
• Develop multimodal retrieval pipelines combining vision-language models for OCR, deep document search, and end-to-end claim attribution.
• Engineer production eval suites and feedback loops that serve as the definitive gate for whether harness modifications and model updates actually improve output quality.
• Collaborate closely with founders and private equity domain experts to turn complex, high-stakes diligence workflows into reliable programmatic systems.

What You Bring

• 5+ years building and scaling production backend services at a venture-backed startup or top-tier tech company, with full ownership of AI-powered systems.
• Proven track record taking LLM features, agents, or complex retrieval systems to production, with eval results and operational war stories to prove it.
• Deep expertise in Python and production search/data infrastructure such as OpenSearch, PostgreSQL, FastAPI, and distributed message queues.
• An eval-driven mindset that treats benchmarking, grounded verification, and error analysis as core engineering responsibilities, not an afterthought.