Associate Member of Technical Staff, AI Engineering

About Himaya

Himaya is an AI-native security platform for small and midsize companies. We protect email against phishing, BEC, vendor fraud and malware. We stop sensitive data from leaving through outbound DLP, monitor SaaS apps, cloud accounts and data stores, and keep customers audit ready. Our engine combines large language models with deterministic detection, sandboxing and threat intelligence. Most threats are investigated and handled automatically, so lean IT teams get enterprise grade protection.

We're a small team, so you will ship to production in your first weeks and own real parts of the system.

The role

You will join the team that keeps Himaya's detection accurate and dependable. Your time will be split across three areas:

detection quality

the open source models behind our reasoning and data classification;

the platform those models run on.

What you'll do

Detection hygiene and improvement:

Review false positives and false negatives from live email traffic, find the root cause and fix it. That might be a pattern rule, a scoring threshold, a prompt or a model.

Maintain and extend our detection rules for phishing, BEC, impersonation, malicious links, attachments and QR codes.

Build and maintain evaluation sets and regression tests, so every change to the engine is measured before it ships.

Monitor model cost, latency and accuracy, and propose improvements backed by data.

Models:

Help stand up and maintain open source models for security reasoning, used in threat triage and investigation.

Build and improve data classification models for Workspace Security and DLP. These detect credentials, personal data, financial data and regulated data across email, SaaS apps and data stores.

Run experiments: prompt design, fine tuning, distillation and model comparisons. Turn the results into production changes.

Keep model serving reliable, with fallbacks, timeouts and observability.

Platform engineering:

Contribute to our Python and FastAPI backend, our workers and our integrations with Google Workspace and Microsoft 365.

Work with Postgres, Redis and containerized services on Azure and AWS.

Improve CI, tests and deployment tooling, and help keep production healthy.

What we're looking for:

A recent graduate in Computer Science, AI or a related field, or up to 1 year of work experience.

Hands-on experience with machine learning or LLMs: building with model APIs, running open source models such as Llama, Qwen or Mistral, fine tuning, or building evaluation pipelines. University projects, research, internships and open source work all count.

Strong Python. You write clean code, and you write tests for it.

Comfortable with SQL, Git and the Linux command line.

Curious about security, and careful by nature. In this job, a wrong verdict can either put an employee at risk or hold back a legitimate email.

Clear written communication. You can explain why a change is right using evidence.

Nice to have

Experience with inference servers such as vLLM, TGI or Ollama, or with cloud model platforms such as AWS Bedrock and Azure AI Foundary.

Experience with text classification, named entity recognition or PII detection.

An understanding of email authentication (SPF, DKIM, DMARC) or phishing techniques.

Docker, any cloud platform, or a background in CTFs and security research.