AI Systems Engineer (Backend + GenAI Infrastructure)

What Youll Work On

• Build Scalable Backend Services: Design and implement robust APIs (FastAPI) that handle multi-turn conversations, manage complex state, and persist execution artifacts (plans, code, logs) to a permanent database layer
• Asynchronous Agent Orchestration: Architect the communication layer between front-end interfaces and backend agents, implementing async patterns (WebSockets, SSE, Queues) to handle long-running tasks
• Cloud & Security: Deploy secure, horizontally scalable infrastructure on platforms like Vercel, Google Vertex AI, or AWS
• Handle authentication, rate limiting, and secure execution sandboxes
• Knowledge Management Support: Build the "Knowledge Studio" backendAPIs and storage schemas that allow users to manage prompts, documentation, and domain concepts dynamically

AI Engineering & ML Ops (30%)

• Model Orchestration: Deploy and manage Small Language Models (SLMs) on the cloud for specialized implementation subtasks (e g , ambiguity resolution), optimizing for latency and cost
• RAG & Context Engineering: Implement retrieval pipelines that dynamically fetch the right context for the agent, ensuring high relevance and low hallucination
• Fine-tuning & Evaluation: Manage pipelines for fine-tuning models on specific domains (coding, tool selection) and running automated evaluations of agent performance

What Were Looking For

Technical Skills

• System Design: Strong proficiency in Python (FastAPI/Django) and experience designing schema for complex applications (PostgreSQL, NoSQL)
• Cloud Native: Experience deploying and scaling on modern cloud platforms (GCP Vertex, AWS, Vercel) and using containerization (Docker/Kubernetes)
• AI Integration: Practical experience integrating LLMs/SLMs via APIs and orchestrating workflows (LangChain, LlamaIndex, or custom orchestration)
• Async Patterns: Understanding of how to handle long-running jobs (Celery, Redis, Queues) and real-time frontend updates

Soft Skills & Mindset

• Engineering First: You treat AI agents as software systems that need to be tested, debugged, and versioned
• Versatility: You are comfortable jumping from optimizing a SQL query to finetuning a Llama-3 model
• Problem Structuring: You can take a high-level goal like "handle ambiguity" and break it down into a technical workflow involving specific model calls and UI interaction states