Site Reliability Engineer, Unstructured.io
+ Equity
Data prepping for Large Language Models
Open for applications
Data prepping for Large Language Models
21-100 employees
Open for applications
+ Equity
21-100 employees
Company mission
To give organizations access to all of their data.
Role
Who you are
- 4+ years of SRE, platform engineering, or infrastructure engineering in a production Kubernetes environment
- Deep operational knowledge of Kubernetes. You've written HPA configs, KEDA ScaledObjects, PodDisruptionBudgets, preStop hooks, and PriorityClasses; you understand pod lifecycle and scheduler behavior
- Demonstrated experience diagnosing and resolving real production performance issues: resource saturation, timeout failures, scheduling problems, graceful shutdown gaps
- Enough Python or Go to read service code, trace a bug to root cause, and write a targeted fix
What the job involves
- The infra team is small, technically deep, and owns the full stack from cloud provisioning and k8s operators to workflow orchestration and core services. We ship frequently and operate at a scale that makes reliability a first-class engineering problem
- As part of the Infra team, you'll work on the reliability and performance of our platform end-to-end.
- You'll work closely with infra and product engineers to instrument systems, establish SLOs, tune autoscaling, and drive incident process maturity.
- The work is technical and hands-on: you'll write code, dig into k8s internals, and hold yourself accountable to positive production outcomes
- Own production reliability across our Knative, KEDA, and Kubernetes-based document processing platform. Proactively detect degradation, diagnose root causes, and ship fixes
- Work on observability: end-to-end tracing, latency SLOs, capacity dashboards, and alerting that finds problems before customers do
- Load testing and capacity planning: establish throughput benchmarks, detect performance regressions before they reach production
- Support fleet operations: contribute to the safe, automated upgrade process for our growing fleet of production systems
Our take
Unstructured.io is tackling a prevalent challenge in AI adoption - the difficulty of integrating complex, enterprise-sized data with Large Language Models (LLMs). The company's platform provides a solution by extracting and staging all data, even that which is in difficult-to-use formats, and making it compatible with LLMs.
Thus, the startup streamlines the time-consuming data preparation process, providing tools for otherwise complex tasks like removing unwanted elements, text concatenation, and Optical Character Recognition (OCR). It also tailors processing pipelines for specific document types, addressing a critical bottleneck in AI development workflows.
In enabling the seamless integration of LLMs with enterprise data, Unstructured's platform has the potential to unlock substantial productivity gains. As of March 2024, its open-source library has been download more than 6M times and by 2025, the total user base has grown significantly, with over 65,000 organizations now leveraging Unstructured tools. The company continues to develop its solution to the data-preparation bottleneck, a crucial component in meeting the evolving demands of enterprise-scale LLM deployments.
Steph
Company Specialist at Welcome to the Jungle
Company
Funding (last 2 of 3 rounds)
Mar 2024
$40m
SERIES B
Jul 2023
$20m
SERIES A
Total funding: $65m
Company HQ
SoMa, San Francisco, CA
Leadership
Brian S. Raymond
(Founder & CEO)Previously worked in Investment Banking at Harris Williams & Co. Prior to this they worked at The White House in the National Security Council and as an Intelligence Officer at the Central Intelligence Agency.
Share this job
View 3 more jobs at Unstructured.io