Software Engineer, AI Platform

Fluency builds a platform that captures how work actually happens inside large organizations, measures productivity and process conformance, and analyzes where AI can do the work. The system captures observable work data across tools and systems, structures it into a model of how work runs, and uses it to measure productivity, check process conformance, and identify where AI changes the work.

 Customers include CVS Health, Aon, and PVH. Fluency is deployed across Fortune 500 organizations where the scale, reliability, and cost requirements are real.

The Opportunity

Fluency is looking for a Software Engineer to own the data platform, ETL pipelines, and agent infrastructure that everything else at the company runs on. This is the platform layer that makes Fluency's AI work reliable, observable, and usable in production — moving data through LLMs, transforming agent outputs into structured downstream data, running jobs reliably, and keeping the system fast and cost-efficient as the company scales.

 This is an early-stage role that requires balancing reliability with iteration speed. You will build the platform, keep it running, and make tradeoffs as priorities shift. On-call participation is part of the role. If you want to own production systems at a company where the infrastructure decisions you make are felt across every AI feature, this is that role.

 This role is not a fit if you want hybrid or remote, are not comfortable with rapid iteration, have not owned production systems or operated production pipelines, do not want to be on-call, or need requirements locked down before you can move.

What You'll Do

• •      Own and evolve the data platform that powers every job across the company


• •      Run the LLM ETL pipeline including ingestion, transformation, enrichment, and storage of LLM-driven data


• •      Build agent transformation infrastructure that converts agent outputs into structured, queryable downstream data


• •      Improve reliability, throughput, and cost of LLM-driven jobs in production


• •      Build observability and tooling so the team can debug and iterate quickly


• •      Partner with AI engineers to expose new platform capabilities and shape the interfaces they build on


• •      Participate in on-call rotation and incident response

You Should Have

• •      Strong Python engineering experience supporting production systems using FastAPI or similar


• •      Experience building or maintaining production pipelines that handle non-trivial volume, retries, backfills, and failure recovery


• •      Hands-on experience with a data orchestrator such as Dagster, Airflow, Prefect, or Temporal, and transformation tooling such as dbt


• •      Comfort with PostgreSQL at scale including schema design, multi-schema setups, and migrations


• •      Comfort with AWS infrastructure including ECS, Lambda, SQS, Step Functions, RDS, and S3, and IaC using Terraform or Terragrunt


• •      Familiarity with LLM APIs and the operational realities of LLM-based systems including latency, cost, retries, structured output, and failure modes

Nice to Have

• •      Experience with distributed compute for Python workloads using Anyscale Ray, Dask, or Spark


• •      Experience with Polars and Pandas for data processing


• •      Familiarity with Datadog for observability, metrics, and tracing


• •      Cost optimization experience for LLM workloads


• •      Familiarity with pgvector or other vector stores


• •      Multi-region AWS deployment experience


• •      Some TypeScript/Node experience, since parts of the platform live there