BRAHMA Careers

• Design and build scalable data pipelines: Architect and implement robust, automated pipelines for ingesting, processing, and managing non-tabular datasets (video, audio, etc.).
• Own and deliver complex work end-to-end: Take advanced pipeline tickets, epics, and infrastructure work from scoping through to production. Proactively gather requirements, flag risks early, and propose solutions.
• Feature extraction: Collaborate with research scientists to identify and extract features and annotations that elevate our dataset quality.
• Data lake management: Maintain our human-centric data lake, ensuring high data quality, lineage, and accessibility for the research team.
• Infrastructure optimisation: Minimise resource consumption and maximise cost-effectiveness by leveraging tools like Ray and Kubernetes.
• Accelerate with AI-assisted development: Use agentic coding tools (Claude Code, Cursor, Copilot, etc.) as a core part of your workflow to maintain high velocity across prototyping, pipeline development, and code review. Push the team to adopt what works.
• Drive continuous improvement: Proactively identify gaps in tooling, processes, and infrastructure. Build internal tools, propose architectural changes, and bring ideas to the table without being asked.

What You'll Need

• Python expertise: Proficiency in Python, with a focus on writing maintainable, clean,
  production-ready code.
• Data engineering fundamentals: Proven experience building (multi-modal) data pipelines and managing large datasets for research and training.
• Multi-modal experience: Previous exposure to, or a deep interest in, processing and analysing non-tabular data (video, audio, etc).
• Orchestration & Cloud: Experience with tools like Airflow, Dagster, or Spark, and being exposed to hands-on cloud work (primarily GCP).
• Distributed compute: Hands-on experience with distributed compute frameworks (
  Ray, Spark etc).
• Autonomy and ownership: You know how to operate with limited context and ambiguity. Gathering requirements cross-departmentally, challenging what doesn't make sense, unblocking yourself, and making decisions to assess with your manager.
• AI-assisted development: Demonstrated work with agentic coding tools and LLM-driven development workflows. You've gone deep enough to have opinions on what works and what doesn't.
• ML mindset: A data-centric approach to AI, with an understanding of how data processing directly impacts model performance.
• Clear, direct communication: You surface problems and ideas without being asked, challenge requirements constructively across teams, and articulate trade-offs clearly to both technical and non-technical stakeholders. Common sense is non-negotiable.
• Proactive drive: A visible track record of bringing ideas - tooling, process improvements, experiments -rather than waiting for them to appear on a backlog.

Nice to Have

• Practical knowledge of computer vision, audio processing, or FFmpeg-based workflows.
• Experience working with generative models or fine-tuning techniques.