About Spearbit & Cantina:
Founded in 2021 by former Ethereum Foundation Solidity engineers, Spearbit tackles Web3 security challenges. Our founding team built the leading blockchain language and secured the largest smart contract, protecting over $160B in value.
We're building Cantina, the "GitHub for Security", connecting security researchers with projects needing expertise. Our Cantina security platform has powered major competitions and serves the leading projects in Web 3. It currently supports collaborative security reviews, public and private security competitions, bug bounty programs, incident response, and AI code analyzer.
The Opportunity: We are seeking a talented Machine Learning Engineer to drive our efforts in leveraging advanced Large Language Models (LLMs) within Cantina Code. You will play a key role in building intelligent features that enhance code security and collaboration on our platform. We are open to a variety of experience levels including recent graduates.
What you'll do:
Design, prototype, and productionise multi‐agent pipelines (e.g., LangGraph / CrewAI‐style graphs) that collaborate to scan Solidity smart‐contract repositories.
Build and fine‐tune bespoke LLMs and smaller custom models for code understanding (AST‐aware transformers, graph‐neural nets, etc.).
Encode and maintain agentic knowledge — task memories, tool use policies, and shared world‐models — so agents improve autonomously.
Implement concept‐drift & data‐drift monitors that retrain or hot‐swap models when blockchain languages or attacker tactics evolve.
Expose capabilities through FastAPI micro‐services and deploy them on AWS.
Collaborate with security researchers to turn novel attack patterns into detection heuristics and learning signals.
Deploy and manage machine learning models and services on AWS.
Collaborate closely with product managers, data scientists, and engineers to integrate AI solutions seamlessly.
What we’re looking for:
Experience with Python and building ML systems in production.
Practical experience fine‐tuning LLMs (LoRA, QLoRA, DPO/RLHF).
Familiarity with agentic / autonomous‐AI frameworks (LangGraph) and their evaluation challenges.
Solid grounding in software‐2‐software architectures: REST contracts, event‐driven workflows, distributed tracing.
Comfort learning/reading Solidity and reasoning about smart‐contract security primitives.
Benefits
Competitive salary and performance-based compensation opportunities
Opportunity to work in an early-stage startup with a talented and passionate team.
Exposure to high-profile clients in the blockchain and cryptocurrency industry
Comprehensive health, dental and vision benefits.
401k matching program
Join Spearbit and help us build the future of code security!
Top Skills
What We Do
Cantina is the leading security platform connecting top researchers and purpose-built tools to protect critical code. Backed by Spearbit’s expert network, we help blockchain and financial teams scale security with high-impact solutions.
