Senior AI Engineer

Unico Connect is seeking a Senior AI Engineer to architect and deliver production AI systems for enterprise clients across the United States, European Union, and Asia-Pacific markets. This role leads multi-agent orchestration, large-scale Retrieval-Augmented Generation (RAG), evaluation infrastructure, and the engineering practices that take artificial intelligence from proof of concept to compliance-ready production. The position suits engineers with a demonstrated track record of shipping production LLM systems and the technical maturity to debug them when they fail.

• Architect production AI systems including multi-agent orchestration, retrieval pipelines, evaluation infrastructure, and observability, with documented design rationale prior to implementation
• Lead RAG implementations at enterprise scale: chunking strategy, hybrid search, reranking, caching layers, and multi-tenant index architecture
• Design model-routing and tiering logic that balances cost, latency, and quality, including build-versus-buy decisions for self-hosted versus frontier model APIs
• Build evaluation harnesses combining golden test sets, LLM-as-judge pipelines, behavioural metrics, and regression detection that gate production deployments
• Mentor 2-3 AI Engineers; conduct AI-specific code reviews where evaluation scores, not unit test pass rates, define the quality bar
• Drive infrastructure decisions across Google Cloud Vertex AI, AWS Bedrock, Modal, and self-hosted serving platforms based on workload, compliance, and cost constraints
• Partner with the Tech Lead, Product Managers, and Senior Backend Engineers on the AI roadmap; negotiate scope when engineering reality requires it
• Deliver voice AI, computer vision, and multimodal features as project requirements demand; the role is not limited to text-based LLMs
• Lead production incident response when agents misroute tool calls or retrieval pipelines surface poisoned context
• Drive AI engineering standards across the firm: prompt versioning, evaluation coverage thresholds, observability baselines, incident post-mortems
• Contribute 1 open-source patch, technical write-up, or internal evaluation framework improvement each quarter
• Represent Unico Connect's AI engineering capabilities in technical conversations with clients, prospects, and architecture review sessions

• 5+ years of professional software engineering experience, with at least 2 years dedicated to production AI / ML systems
• Multiple production LLM products shipped with measurable business outcomes; able to articulate the metrics moved and architecture trade-offs accepted
• Deep proficiency in Python including FastAPI, async patterns, streaming responses, and typed code with Pydantic at production scale
• Strong fundamentals in vector retrieval, embedding models, hybrid search architectures, and reranking strategies
• Hands-on experience with at least 2 of LangChain, LangGraph, LlamaIndex, Haystack, or comparable custom orchestration frameworks
• Demonstrated track record with evaluation tooling such as LangSmith, Ragas, or Phoenix, with the discipline to catch regressions before users report them
• Practical experience with fine-tuning techniques (LoRA, QLoRA, DPO) and self-hosted model deployment using vLLM, TGI, or comparable serving platforms
• Strong fundamentals in distributed systems and asynchronous backend design
• Excellent written communication; architecture decision records, design documents, and incident post-mortems as a default working medium
• Comfort delivering under regulated industry constraints such as SOC 2, HIPAA, GDPR, and FERPA when project scopes require it
• Bachelor's degree in Computer Science, Artificial Intelligence, or a related engineering field, or equivalent demonstrated experience

• MLOps experience with MLflow, Weights & Biases, Kubeflow, model registries, and drift detection systems
• Voice AI experience including Whisper, ElevenLabs, Deepgram, and real-time streaming pipelines
• Computer vision experience with multimodal LLMs and frameworks such as YOLO and SAM
• On-device AI development using Core ML, MLKit, or llama.cpp for mobile-first products
• Open-source contributions, technical conference presentations, or a published technical blog featuring original work
• Experience leading interview loops for AI engineering hires and calibrating technical assessment rubrics

• Large Language Models:Anthropic Claude, OpenAI GPT, Google Gemini, and open-source families including Llama, Mistral, and Qwen
• Orchestration Frameworks:LangChain, LangGraph, custom orchestrators
• Vector Databases:pgvector as default, with Pinecone, Weaviate, and Qdrant when workload demands
• Backend:Python with FastAPI, TypeScript with Node.js and NestJS, PostgreSQL, Redis
• Cloud Platforms:GCP Vertex AI, AWS Bedrock, self-hosted deployments on GKE and EKS, Modal for one-shot experiments
• Evaluation and Observability:LangSmith, Langfuse, Helicone, custom Grafana dashboards

Production engineering at Unico Connect assumes AI tools form part of the daily workflow rather than an experimental augmentation. For this role specifically:

• Claude Code or Cursor as the default integrated development environment
• LangSmith, Langfuse, or Helicone for production AI observability
• Modal or Replicate for short-cycle model hosting experiments
• OpenAI Whisper for transcription and voice workflows
• Perplexity or Claude for research paper digestion and prior-art investigation