About Fuel Cycle:
Fuel Cycle empowers leading organizations with agile research solutions that deliver decision-ready insights — fast, flexible, and fully integrated. As a market research disruptor, our AI-powered Insights Platform is built for speed, precision, and scale. With cutting-edge tools and seamless audience connectivity, we help brands ditch the guesswork and make smarter, customer-led decisions at lightning speed.
Why work at Fuel Cycle?
Join a high-growth team where curiosity is valued, ownership is encouraged, and your work drives real-world impact. Whether you’re based at our Los Angeles HQ, New York City hub, working remotely across the U.S., or part of our global team in India, you’ll help shape the future of decision intelligence for some of the world’s most iconic brands.
Overview:
Build real AI agents on a production platform, with senior engineers working alongside you. You will work in Python and FastAPI, build tool-calling agents on AWS Bedrock and AgentCore using the Strands Agents framework, and own technical decisions within the problems you take on, with your scope growing as you do. This role is fully in-office in Los Angeles because our mentorship model depends on it: pair programming, in-person code review, and a mentor a few desks away.
We hire for fundamentals, curiosity, and initiative. This is an early-career role. Engineers new to the field are welcome, and so are engineers with a year or two of experience who want to go deeper into AI and agentic systems. You do not need to know our stack on day one.
What You’ll Work On:
- Design, build, and improve AI agents, deciding how tool-calling loops, model configuration, and multi-step workflows are structured using Strands Agents and AWS Bedrock
- Build and integrate MCP tools and servers, and learn how tool names and descriptions shape what an LLM does
- Deploy and run agents on Bedrock AgentCore, working with the platform team and your mentor
- Build RAG pipelines over pgvector or FAISS, choosing chunking, embedding, and retrieval approaches based on what the results show
- Design, test, and iterate on prompts; evaluate answer quality, latency, and token cost, and decide which tradeoffs to make; learn how guardrails are applied in production paths
- Evaluate models, tools, and approaches, and recommend which to use, with your reasoning documented
- Ship features and bug fixes in FastAPI services (REST and WebSocket APIs backed by PostgreSQL and Redis), deciding how to design, implement, and test them
- Contribute to background jobs and queue consumers so agent work runs outside the API request path
- Apply our architectural standards, use your judgment on design choices within them, help keep design documents current, and flag where the standards should change
- Submit pull requests, respond to review feedback, and review teammates’ code as you grow
What You Bring:
- Solid Python from coursework, internships, or work: functions, classes, type hints, error handling, testing basics
- Data structures and algorithms, with awareness of time and space complexity
- SQL fundamentals (joins, aggregations, indexes, schema design); PostgreSQL exposure preferred
- REST and HTTP concepts, and how a web API is structured
- Git and pull-request workflow
- Clear communication, comfort asking questions, and a habit of picking up new tools and asking for help when you’re stuck
Preferred:
- An agent project built with Strands Agents, LangChain, the OpenAI SDK, or a similar framework (coursework, personal, or professional), including how an LLM chooses and calls tools
- Model Context Protocol (MCP): building or using MCP tools and servers
- AWS Bedrock or Bedrock AgentCore, or AWS basics such as S3 and SQS
- RAG: chunking, embeddings, and vector stores
- Prompt engineering: writing, testing, and iterating on prompts
- FastAPI or a similar framework such as Flask or Django, SQLAlchemy or another ORM, and a willingness to learn async Python
Experience:
- Bachelor’s or…