Staff Forward Deployed Engineer

Afresh Technologies; New York (NY); Hybrid; USD 150,000 - 230,000; Full time

Benefits offered by this job

Comprehensive health plans

Generous parental leave

Equity packages

401k matching

Flexible vacation policy

Professional development program

Work from home stipend

Job summary

Afresh Technologies is seeking a senior Forward Deployed AI Engineer to join a single team delivering AI into enterprise grocery customers and building the supporting platform. You’ll split time between field deployments and platform work, embedding with customer data and engineering teams, then turning those insights into reusable tooling for the team.

You will partner with account leads and customer technical teams to scope, architect data sources, and productionize pipelines and LLM-powered

Qualifications

• 5+ years building production software and data systems.
• Customer-facing: work with a customer’s engineers and data teams.
• Genuine AI/LLM depth — real systems with retrieval/RAG and tool-use.
• Knowledge graphs, ontologies, or semantic layers in production; graph and vector stores.
• Experience in grocery, retail, or supply chain data domains.
• Prior forward-deployed, solutions, or implementation engineering — or early-stage startup experience.

Responsibilities

• Scope and architect data sources, architecture, and path to production with customers.
• Embed with the customer’s data and engineering teams; integrate into their cloud and data platform; build production-grade pipelines.
• Design and ship LLM- and agent-powered systems with retrieval, agentic workflows, data-quality and analytics agents.
• Harden the platform: knowledge grounding layer and retrieval; reusable building blocks across customers.
• Build evals, tracing, and tooling to measure quality and ship faster on the next customer.
• Own the flywheel: field learnings flow into the platform and platform improvements benefit the next customer.

Skills

Production software

Data systems

Customer-facing

LLM depth

MLOps

Tools

LangGraph

MCP

Databricks

BigQuery

Snowflake

Pinecone

Weaviate

Job description

• As a Forward Deployed AI Engineer, you’re part of a single team that both delivers Afresh’s AI into enterprise grocery customers and builds the platform that makes that delivery fast. You’ll spend dedicated time in the field — embedded with a customer, integrating into their data, shipping AI systems on top of it — and dedicated time on the platform, turning what you just learned into reusable tooling the whole team deploys next. You build the house you live in
• Afresh leads the customer relationship and direction; you and a small team bring the technical firepower — scope and architect the work with the customer, then build it. Because you also own the platform underneath, the rough edges you hit in the field become the things you fix at the root
• This is senior, hands-on, 0-to-1 work in a space with no playbook
• In the field (forward deployed)
• Partner with Afresh’s account lead and the customer’s technical teams to scope and architect the work — the data sources, the architecture, and the path to production
• Embed with the customer’s data and engineering teams (remote and on-site); integrate into their cloud and data platform; build production-grade pipelines and model messy enterprise data into trustworthy data products
• Design and ship LLM- and agent-powered systems on that data — retrieval, agentic workflows, data-quality and analytics agents — reliable enough to run in production, not just to demo
• On the platform (building the house you live in)
• Harden what works in the field into the shared platform: the knowledge and grounding layer (knowledge graph, ontology, and retrieval) that makes grocery data usable by LLMs, the agent frameworks, and the serving infrastructure
• Build the evals, tracing, and tooling that let the team measure quality — accuracy, hallucination rate, latency, cost — and ship faster on the next customer
• Build for leverage: clean interfaces and reusable building blocks, not one-off per-customer code
• Across both
• Own the flywheel: field learnings flow straight into the platform, and platform improvements show up at the next customer

Benefits

• Comprehensive health plans
• Generous parental leave
• Equity packages
• 401k matching
• Flexible vacation policy
• Professional development program
• Work from home stipend

An architect’s instinct: you can take an ambiguous problem and a messy data landscape, design a clean and workable solution, and then build itReal data-engineering depth: building and operating data pipelines, modeling messy enterprise data, and working in a modern cloud data platform (Databricks, BigQuery, Snowflake, or similar)A bias toward ownership and momentum, and comfort traveling to customer sites regularly (~10-20%)We encourage all highly-qualified candidates to apply, even if they do not fulfill all the listed criteriaRange across both modes — you genuinely like being in front of customers and going heads-down to build reusable infrastructure, and you can switch between them without one suffering. This is the role’s defining trait5+ years building production software and data systems, with strong, production-grade codeCustomer-facing comfort: you work well with a customer’s engineers and data teams — running working sessions, explaining your thinking, and earning trust through what you deliverGenuine AI/LLM depth — you’ve built real systems with LLMs and agents (retrieval/RAG, tool-use) and you evaluate quality rather than eyeball itKnowledge graphs, ontologies, or semantic layers in production; graph and vector stores (pgvector, Pinecone, Weaviate) and hybrid searchExperience in grocery, retail, or supply chain data domainsMCP or similar tool/context protocols; agent frameworks (e.g., LangGraph); MLOps, model serving, and observability for LLM systemsPrior forward-deployed, solutions, or implementation engineering — or early-stage startup experience navigating rapid customer expansion