About the role
What you will do
— Design and build LLM-based extraction for fund documents (capital account statements, quarterly reports, NAV updates) with high, measurable accuracy
— Build agents and automations for the first priority use cases: NAV extraction into the portfolio system, an automated monthly intelligence brief, and a company-level portfolio exposure map across investment mandates
— Integrate models through Microsoft Foundry and self-hosted models in the client’s data center, with failover to Azure
— Build evaluation sets and regression tests so extraction quality is tracked with every change
— Work with the Data Engineer on the Fabric data model so extracted data is reusable across use cases
— Integrate with business systems (portfolio management, SharePoint, Outlook/M365, finance and cap table tools) via APIs
— Contribute to architecture discussions with the client’s lead engineer and document key decisions
— Ensure security and data privacy for sensitive financial data
What we expect
— Hands-on experience with Azure (Azure OpenAI, Functions, Container Apps/AKS, Key Vault, Entra ID)
— Practical experience with Microsoft Foundry (Azure AI Foundry): model catalog, deployments, prompt flow / agent service, evaluations
— Working experience with Microsoft Fabric (Lakehouse, OneLake, notebooks) as a data source for AI applications and agents
— Proven experience building document intelligence pipelines that extract structured data from PDFs, emails and reports (Azure Document Intelligence, OCR, LLM-based extraction with schemas and validation)
— Experience with RAG and agent frameworks (Semantic Kernel, LangGraph, LlamaIndex or similar), embeddings and vector search (Azure AI Search or similar)
— Experience with LLM evaluation and quality control: extraction accuracy metrics, guardrails, human-in-the-loop review, cost and latency monitoring
— Ability to work with both commercial (Claude, GPT) and open-weight models (e.g. Qwen) and choose the right one for each task
— Strong communication skills; comfortable working directly with technical leads and business users
— Upper-Intermediate or higher level of English
As a plus
— Domain experience in investment management, private equity / VC fund administration, or fintech (NAV, capital calls, distributions, look-through exposure)
— Experience with crypto / digital asset data sources
— Experience hosting open-weight models (vLLM, Ollama, TGI) on your own GPU infrastructure
— Experience with Copilot Studio, Microsoft Graph API and SharePoint integrations
— Experience with MCP (Model Context Protocol) servers and Claude Enterprise integrations
— Microsoft certifications (AI-102, DP-600)
What you get with Impressit
We are passionate about everything we do — and we back the people who do it.
Ready to join the team?
Apply for Senior AI Engineer (Microsoft AI stack) — or browse the rest of our open roles.