Technisch engineer.

About Ravian AI Ravian AI is building device-native AI systems that can think, decide, and act on behalf of users. Our platform goes beyond web-based agents, enabling true end-to-end automation directly on devices. We’re working with enterprises and consumers to unlock productivity and decision intelligence at scale. You will design and build production-grade, multi-agent AI systems and the backends that power them. You’ll work across FastAPI services, WebSocket event streams, orchestration logic (Autogen/LangGraph), and device execution layers to deliver agents that complete real tasks—autonomously, safely, and repeatably. Develop high-performance APIs with FastAPI, implement WebSockets for streaming I/O, and own service-level reliability (timeouts, retries, backoff, circuit breakers). handle tool calling, memory, state, guardrails, and safe-actions. Productionize AI: Turn models + prompts into idempotent workflows with input validation, schema contracts, evaluation harnesses, and offline/online metrics. Ship deep tracing, structured logs, and red-teaming hooks. Build guardrails to prevent bad actions and ensure reversible operations. Performance engineering: Reduce latency, control costs, and design fallbacks (models, tools, or policies) for robustness under failure. Quality bar: 3–5 years of software/AI engineering in production environments. ~ Strong backend skills with FastAPI (or similar), WebSockets, async Python, and at least one cloud(AWS/Azure/GCP). ~ Proven record of having built and shipped AI projects/products used by real users (internal or external). ~ Solid understanding of LLM tooling (prompting, tool-use, RAG, evals, safety/guardrails). ~ Comfortable with data models, queues, caches, relational/NoSQL storage, and containerization. Nice-to-Have (Bonus) Experience with device control (browser automation, OS-level actions, mobile/desktop app automation). Experience with observability stacks (OpenTelemetry, Prometheus/Grafana, ELK). Freshers. Applicants who haven’t shipped AI systems (toy demos/ usual projects don’t count). Candidates whose background is only classical ML model training without production agentic experience or have deep learning experience. Python, FastAPI, WebSockets, Autogen, LangGraph, Celery/Queues, SQL/NoSQL, Docker, AWS (Lambda/ECS/S3), OpenTelemetry, CI/CD. Intro call (30 mins): What you’ve shipped. ai and # mention both Subject: Application — AI Engineer — Your Name A production AI system you shipped, your role, scale, and business impact. Clarify if you identify as Data Scientist or ML Engineer with production multi-agent systems experience. Note: If your background is limited to training classical ML models without production agentic work, please do not apply. Compensation ₹8–12 LPA (INR) based on experience and fit. Potential performance-based upside. NOTE : Please ensure you must have a clear understanding of Ravian AI, its mission, and the applications it builds before applying or engaging further. Familiarity with the company’s device-native AI systems, which enable end-to-end automation and decision intelligence, is essential for alignment and meaningful contribution .