Senior Software Engineer - Applied AI / Cloud

Opentrons•Long Island City, NY

3h•Hybrid

About The Position

Opentrons Labworks, Inc. is seeking a Senior Applied AI Engineer (SWE3) to contribute to the development of AI-powered features within OpentronsAI, a platform designed to enhance laboratory automation using Large Language Models. This role involves building and maintaining AWS-based microservices and asynchronous workflows for protocol generation, integrating LLM calls and knowledge retrieval systems into production SaaS infrastructure, and ensuring the reliability and safety of these systems at scale. The engineer will focus on the ML/AI systems layer, including natural language processing for understanding scientific input, classifiers for request routing and scoring, and embedding infrastructure for grounding model output in protocol knowledge. A strong emphasis will be placed on evaluation, testing, benchmarking, and implementing cost and safety measures like token spend tracking and guardrails to ensure system trustworthiness. The role requires applying rigorous engineering principles to aspects such as observability, cost control, and safe failure modes.

Requirements

Bachelor’s degree in Computer Science or similar field, or equivalent certifications and experience.
5+ years of professional software engineering experience.
At least 3+ years spent building, deploying, and maintaining production AI/ML or LLM-powered applications.
Experience shipping and supporting a SaaS product through a real launch, including operational work.
Solid hands-on experience working within cloud microservice architectures (AWS strongly preferred), including deploying, scaling, and operating microservices in production.
Systems design experience: API design, service boundaries, data flow, and tradeoffs in scaling distributed systems.
Hands-on experience building production AI or machine learning systems from prototype to production (beyond integrating third-party APIs), including NLP and classifier work.
Experience building RAG pipelines: chunking, embedding/indexing, and retrieval tuning.
Experience designing evaluation frameworks or rubrics to measure AI output quality.
Practical understanding of AI safety and cost discipline: guardrails, validation, and token spend monitoring.
Proficiency in React/TypeScript and Python/FastAPI, with experience designing RESTful APIs and full-stack debugging.
Familiarity with Git workflows and CI/CD practices.
Strong communicator who can own features end-to-end, balance speed with code quality, and thrive in a fast-paced startup environment.
Solid software engineering fundamentals: testing, debugging, and writing maintainable code.
Comfort with ambiguity and iterating quickly based on evaluation results and feedback.

Responsibilities

Design, build, and operate AWS-based microservices and asynchronous workflows for protocol generation, including API design, service boundaries, message-driven communication, and integration with the SaaS platform.
Apply systems design principles to integrate AI components into the broader architecture, considering request routing, service-to-service communication, data flow, and scaling under production load.
Build the ML/AI systems layer, including NLP pipelines for parsing scientist input, classifiers for routing and scoring requests, and embedding infrastructure for grounding model output in protocol knowledge.
Build and iterate on knowledge retrieval and retrieval-augmented generation (RAG) pipelines, including chunking, embedding/indexing, and retrieval tuning.
Design and apply evaluation frameworks and rubrics to benchmark model and application output, and use them to drive iteration before issues reach production.
Build cost and safety instrumentation, including token spend tracking, guardrails, and validation, to ensure AI services are trustworthy and sustainable at scale.
Instrument services for observability (logging, metrics, tracing) and build monitoring to catch failures before customers do.
Support features post-launch, including on-call debugging, incident response, and iterating on the architecture based on real production usage.