Security Engineer

We're hiring a Security Engineer to build and operate the controls that protect Allium's infrastructure, applications, employee systems, and use of AI. You'll work closely with Engineering and the person leading our security program to investigate threats, implement practical fixes, and automate security work.

As AI agents gain access to code, data, and tools, security needs to cover both human and machine activity. You'll help secure how agents operate while strengthening our defenses against automated attacks, credential theft, and compromised software dependencies. This is a hands-on engineering role: you'll write code, test controls, investigate real signals, and take findings through remediation.

We invite people of all backgrounds (https://www.allium.so/about). We have engineers who learnt coding much later in life, who learnt coding on the side, we have engineers who are still in school and we also have engineers who went to the top schools (CMU, Stanford, UIUC, UPenn, Oxford, NUS, Cornell), all are welcome if one comes in with a curious mind and an infectious work ethic.

[https://app.ashbyhq.com/api/images/user-content/2f62de56-0dd0-4792-90f8-069c5cb51068/68e40347-6f06-4fcd-a99a-68f539058433/allium%20team%20montage.jpeg]

[https://app.ashbyhq.com/api/images/user-content/2f62de56-0dd0-4792-90f8-069c5cb51068/1d38ca35-e23a-4c10-9a8f-8a79993c8cb3/image.png]

• Secure cloud infrastructure and access. Implement least privilege, phishing-resistant authentication, short-lived credentials, controlled privileged access, and reliable provisioning and deprovisioning. Improve cloud configuration, managed credential storage, and network protections with service owners. Improve secrets scanning and prevent credentials from entering source control.
• Secure AI agents and developer tools. Help implement permissions, execution isolation, network restrictions, and audit trails for coding assistants, agents, and tool integrations, including MCP. Test how untrusted content could trigger prompt injection, tool misuse, or unintended data access, and build safeguards at the point where actions execute.
• Build detection and response. Correlate identity, cloud, workload, and CI/CD telemetry to detect credential theft, unusual access, unauthorized agent actions, suspicious software changes, and data exfiltration. Investigate attacks across systems, reduce alert noise, and test response procedures with system owners, including session revocation, exposed-credential rotation, and removal of unauthorized access.
• Reduce software supply-chain risk. Assess dependencies, CI/CD workflows, third-party integrations, and developer environments. Partner with engineers to validate findings, protect build and deployment credentials, and remediate vulnerabilities.
• Move quickly on relevant threats. Maintain visibility into internet-facing services, including developer tools and management interfaces. Evaluate new vulnerabilities and active exploitation against our actual exposure. Help patch, isolate, revoke credentials, or apply compensating controls with service owners, then verify that the response worked.
• Automate and document. Use code and APIs to improve access reviews, configuration checks, evidence collection, and remediation tracking. Use AI thoughtfully for security work, validating its output and protecting sensitive data.
• Support employee security and enterprise assurance. Deploy, validate, and maintain endpoint protections and device security baselines, including prevention settings, tamper protection, encryption, and timely updates. Work with the owner of employee IT on identity and SaaS controls. Provide clear technical evidence for audits and customer reviews, and maintain useful runbooks.
• WHAT WE'RE LOOKING FOR
• Typically 4+ years of relevant security, infrastructure, or software engineering experience, with direct responsibility for implementing security controls.
• Hands-on experience with a major cloud platform and a strong foundation in identity, networking, operating systems, and access control.
• Depth in at least two relevant areas, such as cloud security, detection and response, application security, vulnerability management, or corporate security.
• Experience investigating technical problems from logs and system behavior, testing hypotheses, and following issues through to resolution.

• Permissionless nature leads to fragmentation of meaning
• Decentralization leads to too many standards https://xkcd.com/927/
• Immutability leads to exponential data + query infrastructure complexity
• Neutrality means that no one is accountable for interpretations

Blockchain data is public. It is not usable at institutional scale. Despite being open, blockchain data is fragmented, hard to interpret, and painful to maintain. Even a simple question like “Who are the largest Ethereum token holders over time?” can require running nodes, ingesting full chain history, decoding contracts, cleaning edge cases, and writing complex SQL.

BLOCKCHAINS ARE OPTIMIZED FOR WRITES, NOT READS.

They are built for consensus and execution, not searchability, standardization, or financial interpretation. Blockchains are computers, not databases https://x.com/Tim_Roughgarden/status/1512847113599729668. Every protocol defines its own schema. The same economic action can be encoded in dozens of different ways. The result:

• Fragmented standards
• Exponential complexity
• No accountability for interpretation
• Events without economic meaning

Finance cannot operate on that, it needs an effective system of record.

Allium ingests, verifies, and standardizes data across 140+ blockchains and 30+ petabytes of history. We close four structural gaps that prevent blockchains from becoming systems of record:

• Semantic Gap: Translating raw events into financial concepts like payments, trades, deposits, and staking income
• Standardization Gap: Mapping thousands of protocols into a single canonical cross chain schema
• Infrastructure Gap: Read optimized, globally distributed data at web scale
• Accountability Gap: Auditable methodology, SLAs, and SOC 1 and 2 compliance

The result is a neutral, canonical data layer institutions can build on with confidence.

Stablecoins, tokenized assets, trading, staking, and lending are growing rapidly. Institutions need a trusted source of truth for onchain financial activity, just as they rely on Bloomberg or DTCC in traditional markets. Raw blockchains cannot serve that role.

As AI agents begin transacting autonomously https://x.com/AlliumLabs/status/2019810793324482901?s=20, the requirement becomes even stricter. Agents cannot reason over raw event logs. They need structured data, attribution, condition checks, and auditability.

Allium is the read layer that makes onchain finance usable for humans and machines.

• Finance, Accounting, and Risk Teams They need reliable, audit grade answers. They rely on Allium for financial reporting, reconciliation, compliance, risk monitoring, and defensible metrics that can stand up to auditors and regulators.
• Engineers and Product Teams They need low latency, production ready infrastructure. They use Allium to power wallets, trading systems, payment rails, staking infrastructure, and real time applications that cannot break.
• Strategy, Research, and Executive Teams They need clarity and insight. They use Allium to understand ecosystem economics, market structure, user behavior, competitive dynamics, and where capital is flowing onchain.

And of course...agents 🦞. Our customers and users include Visa, Stripe, G-SIB Banks, Big 4 Accounting firms, BCG, Coinbase, Phantom, Uniswap and cited by the Federal Reserve.

• Medical, Dental, Vision, Life and AD&D insurance - US folks get 100% coverage for Gold plans, 80% for dependents
• Ownership – Meaningful early-stage equity. Every full-time employee receives a stock option grant so you can share directly in Allium's upside
• Time off – Flexible, trust-based paid time off. We encourage people to take the rest they need. In every country we hire, our policies meet or exceed local statutory vacation requirements
• Retirement – Company sponsored 401(k) plan in US so you can save pre-tax for the long term

All applicants have to answer this pop quiz: "What is an Allium? What is your favorite Allium? https://www.allium.so/post/how-allium-got-its-name". Bonus points for the right pronunciation.