AI engineer (remote, 4 months)

About MincaAI

MincaAI is an AI-native software company transforming the insurance industry. We don't build chatbots or isolated tools. We redesign and automate entire workflows — from intake to. decision-making — by combining LLMs with deterministic systems where each belongs.

Our platform connects directly to enterprise core systems to replace manual, fragmented processes with

scalable, AI-driven operations.

The Role

We are looking for an AI Engineer with strong AI/ML engineering fundamentals and hands-on experience

building production AI systems that handle messy, real-world data at scale.

You will design, build, and deploy extraction and retrieval pipelines that turn unstructured inputs

into structured, trustworthy outputs powering live business workflows.

What You'll Do

- Design extraction pipelines that pull structured data from unstructured documents (Excel, email, PDF)

with inconsistent layouts, missing fields, and adversarial edge cases

- Build hybrid matching and entity resolution systems combining embeddings, BM25/lexical search,

rule-based filters, and LLM-based re-ranking

- Implement multi-stage ranking and fusion pipelines, then tune confidence thresholds, fallback logic,

and human-in-the-loop handoffs

- Decide where LLMs earn their cost vs. where deterministic code, regex, or classical ML wins on

latency, cost, and reliability

- Improve accuracy, calibration, and throughput of AI workflows in production — measure, ablate, ship

- Integrate AI systems with existing enterprise infrastructures (APIs, databases, core systems)

- Design scalable backend architectures for AI applications: queues, workers, caching, vector stores

- Work closely with our PM and clients to translate business problems into concrete pipelines and

evaluation metrics

What We're Looking For

- 5+ years of experience in AI and/or ML engineering

-Degree in Computer Science, AI, ML, or similar fields. (Masters is preferred).

- Hands-on production experience with LLM-based extraction, classification, or agentic workflows — not

prototypes, not demos

- Strong judgment on when NOT to use an LLM: comfort with deterministic parsing, rules engines, and

small models when they are the right tool

- Familiarity with vector search and embedding-based retrieval (pgvector, FAISS, Qdrant, or similar)

and hybrid search patterns

- Strong Python backend skills: FastAPI or equivalent, SQLAlchemy/Pydantic, system design, relational

databases

- Experience designing evaluation harnesses, confidence scoring, and threshold tuning for ML/LLM

systems

- Experience working with large international companies is a strong plus

- Fluency in English

- AI-native development mindset:

- At least 1 year of active use of AI coding tools (e.g. Claude Code, Cursor)

- Ability to leverage AI to ship faster, better, and cleaner code

- Comfortable working in a fast-moving, AI-first environment

Our Values

Growth Mindset — You learn fast, iterate, and constantly improve.

Full Ownership — You take responsibility end-to-end. No excuses, no silos.

AI-First, Always — AI is not a feature — it's the foundation of how we build.

Why Join MincaAI

- Fully remote with an international team

- Work on high-impact AI systems deployed in real enterprise environments

- Build products that replace legacy workflows, not just decorate them

If you want to build real AI systems — not demos — and work on extraction, retrieval, and pipeline

problems that actually matter, we should talk.