Associate AI Engineer
Description
The Associate AI Engineer will build and evaluate LLM-powered features for FarmerChat, Digital Green's conversational AI platform serving millions of smallholder farmers. We are seeking an engineer with Python skills and hands-on LLM experience to design, test, and monitor AI systems that work reliably in multilingual, low-resource, and real-world agricultural contexts. You'll rotate across feature development, evaluation testing, and observability depending on sprint priorities-gaining broad exposure across the full AI stack.
In this role, you will write and refine prompts, build agentic workflows using tool-use patterns and the Model Context Protocol (MCP), maintain embedding and vector store pipelines, and work with multimodal data (text, speech, image) to serve farmer-facing advisory tools. You'll define and run automated evaluation frameworks to catch hallucination and bias before models ship to production, and collaborate with senior engineers, product, and field teams to ship reliable AI systems. This is a generalist role with genuine mentorship-your contributions matter from day one, and you'll grow rapidly in a fast-moving, impact-driven team.
Key Responsibilities:
- Build & Integrate LLM Features
Help build features for FarmerChat using LLM provider APIs (Anthropic Claude, OpenAI, Google Gemini) with guidance from senior engineers. - Write and refine prompts, call APIs, and wire up tool-use patterns using provider SDKs and the Model Context Protocol (MCP).
- Test & Evaluate AI Systems
Run and extend automated test suites that check LLM outputs for accuracy, hallucination, and bias using tools like Ragas, DeepEval, or Promptfoo. - Test agent workflows end-to-end and document edge cases, failure modes, and regressions.
- Work with Embeddings, Speech & Vision
Help maintain embedding pipelines and vector stores (Qdrant, pgvector, FAISS) that ground LLM responses with relevant agricultural knowledge. - Support the team's expanding multi-modal work - speech-to-text (Whisper, Deepgram), text-to-speech, and image classification for crop and pest detection.
- Monitor, Debug & Document
Use observability tools (Langfuse, Arize Phoenix, Weights & Biases) to monitor system health and flag issues. - Document your work - prompts, experiments, runbooks - so the team can reproduce and build on it.
Qualification & Skills:
- 1-4 years in software engineering, ML, data science, or a related AI role.
- Bachelor's or Master's in Computer Science, AI/ML, or a related field.
- Solid Python skills - comfortable writing code beyond notebooks.
- Hands-on experience with at least one LLM provider API (Claude, OpenAI, or Gemini).
- Familiarity with prompt engineering basics (few-shot, chain-of-thought, structured output).
- Basic understanding of embeddings and vector stores.
- Git and collaborative development workflows.
Nice to Have:
- Exposure to agentic workflows, MCP, or tool-use patterns.
- Hands-on with speech models (Whisper, Deepgram) or vision models (ViT, CLIP, YOLO).
- Experience with LLM evaluation frameworks (Ragas, DeepEval, Promptfoo).
- Cloud platform basics (AWS Bedrock / SageMaker, GCP Vertex AI, or Azure).
- Fine-tuning experience (LoRA / PEFT) or open-source LLMs (Gemma, LLaMA, Mistral).
- Docker / containerization basics.
What You'll Build:
- Production features for FarmerChat - a real AI product serving millions of users.
- Evaluation pipelines that validate every model update before it ships.
- Multi-modal systems spanning text, speech, and image classification for agricultural use cases.
- Tools and documentation that make the whole AI team more productive.
Skills
Sign in to read and join the discussion.