Senior AI/ML Engineer

Location: US - NYC, Boston

Salary: $200,000 - $280,000

💎 About Nivoda

Nivoda is the operating system for the global jewellery industry. We connect diamond and gemstone suppliers directly with jewellery retailers across more than 70 countries, replacing a slow, offline, relationship-gated way of trading with a single digital marketplace. We're already powering over $300M in annual transactions, inside an industry worth more than $350B globally — and we're still early. Jewellery is one of the last major global industries to go digital, and we're the ones building the rails it will run on.

🤔 Our Mission

Diamond and gemstone trading has run for decades on fragmented networks, manual processes, and who-you-know relationships — which locks smaller and newer businesses out, and leaves even the biggest players losing time and margin to inefficiency. We started Nivoda to fix that: to give any jewellery business, wherever it is and whatever size it is, the same access, pricing transparency, and ease of trading as an industry giant. Our mission is to make buying and selling diamonds and gemstones radically simpler, everywhere in the world.

⭐️ The Discovery Pod

You'd be joining the Discovery Pod 🔍 — the team that decides how hundreds of thousands of diamonds and gemstones actually get found. That means ranking, recommendations, and search input: from the core search and filtering experience, to Contextual Search and a planned conversational assistant, to image-upload search that lets a buyer find a match from a photo. The same ranking and recommendation engine you'll build has to serve both Nivoda's own marketplace and the many individually branded storefronts jewellers run through our SAAS products. We're a small, high-ownership team and this pod is building the AI-powered product itself, not just using AI tools to go faster.

About the Role

We're hiring a Senior AI/ML Engineer for the Discovery Pod to help build and run the ranking, recommendation, and search systems buyers rely on to find the right stone or piece in seconds. This is real information retrieval and applied ML work on production infrastructure — search systems, embedding-based retrieval, learning-to-rank, LLM-powered contextual and conversational search, image embeddings for photo-based search — not configuring a hosted search widget or calling a model API and moving on. You'll own models end to end, from features and training through evaluation, deployment, and monitoring, and you'll be expected to understand how these probabilistic systems behave and fail, because that understanding is the job, not an adjacent skill.

Tech Stack

Search: OpenSearch, fed via CDC from PostgreSQL plus Spark/EMR/Glue ingestion jobs. No vector database anywhere in production — embedding-based retrieval, the ranking model, and image-embedding search are genuinely new builds for whoever takes the Discovery role.

Core DB / warehouse: PostgreSQL (OLTP) → Snowflake (dbt-modeled, medallion raw/curated/presentation layers)

Backend: TypeScript/Node.js, GraphQL API gateway, Nx monorepo + PNPM

Auth: Keycloak · SSO: Okta

CI/CD: Jenkins (“Nivoda Build”) + GitOps for deploys

Cloud: AWS (EMR, Glue, DMS, Kubernetes-based deploys)

Observability: Grafana for logs/traces

What would you be doing?

Core Search Engine

Build and operate real search infrastructure — Elasticsearch/OpenSearch, Vespa, or vector-search databases like pgvector, Pinecone, or Weaviate — to power ranking and multi-modal search across Nivoda’s marketplace

Design multi-tenant, configurable ranking and recommendation systems that serve both Nivoda’s own marketplace and many individually branded jeweller storefronts through Feeds, without forking the codebase

Multi-Modal AI

Build and improve Contextual Search and our planned conversational assistant, using prompt design and retrieval-augmented approaches, with zero-result and tradeoff scenarios handled gracefully rather than met with a confident wrong answer

Build image-to-attribute and image-to-product-match pipelines using image embeddings for our image-upload search work

Build the product around the reality that probabilistic systems can be confidently wrong — designing recovery paths and confirmation steps before high-stakes actions, like generating a manufacturing quote from an interpreted image

Experimentation & Quality

Define and run search- and ranking-specific evaluation — offline ranking metrics, interleaving experiments — to tell whether a ranking change is actually better, not just different

Evaluate and stress-test foundation model and LLM output directly as part of the engineering process, rather than trusting it by default

Production ML

Own the full lifecycle of ranking and recommendation models — features, training, evaluation, deployment, and monitoring — using learning-to-rank, embedding-based retrieval, and collaborative filtering techniques

Design and optimize low-latency, real-time model-serving systems that keep search feeling instant under production load

Use AI coding agents to build ML infrastructure and pipelines, paired with rigorous evaluation harnesses that catch ranking and retrieval bugs a system can hide while looking like it’s working

What do we need from you?

Years of experience: 5-10 years

Information retrieval infrastructure experience — building or operating real search systems (Elasticsearch/OpenSearch, Vespa, or vector-search databases like pgvector, Pinecone, or Weaviate)

Applied ML for ranking and recommendations, end to end — learning-to-rank, embedding-based retrieval, collaborative filtering — with comfort owning features, training, evaluation, deployment, and monitoring, not just calling a model API

NLP and LLM application experience — production experience building on top of language models, including prompt design, retrieval-augmented approaches, and handling zero-result and tradeoff scenarios gracefully

Low-latency serving experience — you've built and optimized real-time model-serving systems under production load, not just trained models in a notebook

Specialized evaluation skills for search and ranking quality, distinct from general growth experimentation — offline ranking metrics, interleaving experiments, and judging whether a ranking change is actually better versus just different

Direct, hands-on experience building with foundation models and LLMs as an engineering material — prompt design, retrieval-augmented approaches, and critically, evaluating and stress-testing model output rather than trusting it by default

Experience designing for graceful failure in probabilistic systems — knowing when a model is confidently wrong and building the product around that reality

Comfort using AI coding agents to build ML infrastructure and pipelines, paired with strong evaluation and testing discipline

Nice to have

Computer vision fundamentals, specifically image embeddings — experience building image-to-attribute or image-to-product-match pipelines

<li