§ Capabilities
AI engineering
Fine-tuning small models for domains with scarce data, and building the agent systems and evaluation harnesses around them.
- LLM fine-tuning (Unsloth, Hugging Face Transformers)
- Qwen, Llama 3.2, Granite, DeepSeek
- OpenAI Agents SDK
- LangChain / LangSmith
- Model Context Protocol (MCP)
- Multi-agent architectures
- Harness-level evals and graders
- Hallucination detection
- Retrieval-augmented generation
- Vector stores
- Azure OpenAI
- Anthropic Claude, OpenAI GPT