AI Engineer

About the job

● Agentic AI & LangGraph ○ 3+ years building production multi-agent / workflow systems with LangGraph (or similar). ○ Supervisor patterns, sub-orchestrators, worker agents, shared state, checkpointing, time-travel debugging. ● AI-assisted Coding Workflow ○ Daily user of AI coding tools (Claude Code, Cursor, VS Code AI extensions, MCP servers). ○ Designs and manages sub-agents, hooks, background tasks, and toolchains. ● Python Backend & APIs ○ 5+ years Python, including async/await and concurrency patterns. ○ Production-grade FastAPI services and API gateways. ○ Celery + Redis (or equivalent) for distributed task queues, background jobs, and long-running AI workloads. ● Multi-LLM Integration, Orchestration & Evals ○ Integrates multiple providers (OpenAI, Anthropic, Google, xAI, etc.) in one system. ○ Strong with prompt design, tool/function calling, structured JSON outputs, streaming responses, routing & fallbacks, guardrails, and cost control. ○ Builds evaluation pipelines: LLM-as-judge, human-in-the-loop evals, automated regression testing for prompts/agents. ○ Familiar with prompt optimization techniques (e.g., GEPA-style generate–evaluate–prompt–adjust loops). ● Web Scraping, Crawling & Agentic Search ○ Hands-on with Firecrawl, Playwright, Selenium (or similar) for crawling and browser automation. ○ Converts web content into LLM-ready formats with robust error handling and de-duplication. ○ Designs agentic search / deep-research workflows using Bing/Azure and similar APIs. ● RAG, Vector Stores & Knowledge Graphs ○ Production use of vector databases (Pinecone or similar) for semantic search and conversation memory. ○ Builds pipelines for skill graphs, user embeddings, content vectors, and retrieval policies. ○ Hybrid search (sparse + dense), chunking strategies, metadata/RBAC, and multimodal embeddings. ● Real-Time Streaming, Caching & Performance ○ Redis for session context, hot-content caching, pre-computed outputs, and rate limiting. ○ SSE/WebSockets for token streaming and real-time UI updates. ○ Designs pre-computation layers and cache hierarchies for multi-tenant AI workloads. ● Data Stores & Analytics ○ PostgreSQL schema design, migrations, and query optimization for transactional workloads. ○ DynamoDB (or similar NoSQL) for high-throughput document/key-value storage. ○ ClickHouse or equivalent columnar store for analytics, event tracking, and BI integrations. ● Multimodal AI, Image Generation & Document Processing ○ Speech pipelines using Whisper (STT) and ElevenLabs (or equivalent TTS) in production. ○ Vision models for screen/UI understanding, OCR, and UI analysis. ○ Robust PDF/image processing with PyMuPDF, OCR stacks, and content normalization. ○ Image generation using Google Gemini 3 Pro ● Full-Stack Product Engineering ○ React 18, Next.js 13/14, and TypeScript for production frontends. ○ TailwindCSS, shadcn/ui, and component-driven design. ○ Builds real-time chat/voice interfaces connected to streaming AI backends. ○ Experience with browser extensions (screen capture, UI highlighting) and modern DevOps (Docker, CI/CD, cloud deployment).

Requirements

• Python
• AI Tools
• FastAPI
• Celery
• Redis
• Web Scraping
• Data Stores

Preferred Technologies

• Python
• AI Tools
• FastAPI
• Celery
• Redis
• Web Scraping
• Data Stores