Overview
Software Engineer II on the ADEX team, owning impersonation detection (brand, lookalike-domain, VIP and employee impersonation) end-to-end from design through launch, operation and continuous improvement.
What you'll do
- Design, build, and operate detection that is core to Abnormal's products, from initial design through rollout, monitoring, and ongoing maintenance.
- Own detection projects end-to-end, including those that begin with a degree of ambiguity: scope loosely defined problems, identify risks, define milestones, and deliver reliably.
- Analyze attacks that get through: pull and study missed-attack data, read messages the way an attacker and an analyst would, identify the underlying pattern, and translate it into detection enhancements or new detection systems.
- Write and tune detection logic using scored signals and attributes, add new signals across the pipeline, and drive changes to launch with a strong focus on minimizing false positives.
- Build and evaluate LLM-based detection agents, and measure precision and recall rigorously with evaluation tooling.
- Surface detections as reusable intelligence that other products and teams across the platform can consume.
- Participate in the on-call rotation for detection surfaces, debug and resolve customer escalations, and feed learnings back into design, observability, and runbooks across regions.
- Leverage AI as a core part of the development loop for code, tests, data analysis, experiments, and documentation, while maintaining strong engineering judgment and validation practices.
- Contribute to team health and culture by documenting heavily, sharing learnings, and giving thoughtful feedback in code and design reviews.
What you'll need
- 3+ years of professional software engineering experience, with a track record of shipping and operating production systems.
- Strong software engineering fundamentals: data structures, algorithms, system design basics, testing, debugging, and clean, maintainable code.
- Strong Python proficiency and comfort learning new languages and frameworks as needed.
- Solid data-analysis instincts; comfortable with SQL and reasoning over large datasets to find signals in noise.
- A detection or adversarial mindset: enjoys thinking like an attacker, reading real attack samples, and asking how to get past a control.
- Genuine fluency with AI-native development; already uses AI coding agents in daily work and is excited to build LLM-powered detection.
- Demonstrated ability to own projects with initial ambiguity: clarify and scope loosely defined requirements, make tradeoffs explicit, deliver on time, and communicate status clearly.
- Excellent written and verbal communication, especially in remote, distributed teams.
- A strong growth mindset and sense of ownership.
Nice to have
- Experience with distributed systems, high-throughput pipelines, or large-scale data stores (e.g., PostgreSQL, DynamoDB, Redis, RocksDB, Kafka, Spark, OpenSearch/Elasticsearch).
- Background in security, threat detection, anti-abuse, fraud detection, or trust and safety, particularly systems processing high volumes of email or communication data.
- Experience with ML or LLM evaluation: precision/recall tradeoffs, eval harnesses, prompt iteration.
- Familiarity with domain and DNS concepts (such as typosquatting and homoglyphs) or with identity and impersonation signals.
- Experience with large-scale data tooling (e.g., Databricks, Spark, Airflow) and distributed pipelines.
- Experience with containerization and orchestration (Docker, Kubernetes) and infrastructure-as-code tooling.
- Familiarity with modern frontend frameworks (e.g., React) for full-stack roles, or with ML/ML Ops for Detection/MLE-focused roles.
- Prior experience in a fast-paced, high-growth startup environment, balancing speed, quality, and ambiguity.
Details
- Hybrid, based in Bangalore, India.
Read the full description and apply on the company’s own careers page.