Applied AI Engineer - Peryx.ai

About the Role

Peryx.ai is building an AI-first intelligence and autonomy platform for renewable-energy operations. We're looking for a hands-on AI builder who has worked on real LLM/ML projects and enjoys learning quickly, experimenting, and turning ideas into working systems.

You'll work closely with the founding team across conversational AI, RAG, AI-assisted workflows, evaluation, automation and production AI systems.

What You'll Work On

• Build and improve LLM-powered product features and conversational AI workflows.
• Develop RAG pipelines over technical and operational knowledge.
• Build structured AI workflows using tool/function calling and schema-constrained outputs.
• Work on AI-assisted data mapping, classification and reasoning workflows.
• Create evaluation datasets, benchmarks and regression tests to measure AI quality and reliability.
• Build feedback loops to continuously improve AI performance.
• Integrate AI services with backend APIs, databases and existing platform services.
• Experiment with different models and techniques to improve domain-specific intelligence, accuracy and efficiency.
• Test AI systems for hallucinations, prompt injection, incorrect tool usage and other failure cases.
• Use AI development tools such as Claude Code, Codex and Cursor to prototype, test and ship features faster.

What We're Looking For

Strong Python fundamentals and hands-on experience building AI/LLM projects are the main requirements. You should be comfortable working with APIs, JSON, Git/GitHub and basic backend development, and have a good understanding of concepts such as LLMs, prompting, RAG, embeddings, vector search, structured outputs, tool/function calling and AI evaluation.

We value people who can independently move through:

Problem → Understand existing system → Build → Test → Evaluate → Improve

Good to Have

• Experience experimenting with open-source or smaller language models
• Familiarity with Hugging Face, PyTorch or PEFT
• Exposure to LoRA / QLoRA, fine-tuning or model adaptation
• Understanding of LLM evaluation, benchmarking and dataset preparation
• Basic knowledge of model inference, quantization or serving
• Experience with LangGraph, Pydantic AI, LangChain, LangFlow or similar frameworks
• Experience with vector databases and agentic workflows
• Familiarity with FastAPI, PostgreSQL, Docker, AWS, Kafka or MQTT
• Understanding of async Python, concurrency or scalable APIs

What Matters Most

We care more about what you have built and how you think than years of experience or certificates.

A strong candidate is curious, learns fast, can debug independently, understands enough software engineering to work inside a real codebase, and knows how to use AI tools to move faster without blindly depending on them.