1. Role Overview CERES AI is building an Agentic AI Platform to operationalize more than a decade of proprietary agricultural data, AI/ML models, and domain expertise. We are seeking a senior-level remote AI/ML Engineer based in Ukraine to design, develop, and deploy multiple AI agents across sales, customer success, operations, and insurance claims workflows.
This is a high-impact, technically demanding role. The engineer will work closely with the product team, and U.S.-based engineers to build production-grade multi-agent systems that ingest real-world data (ERA5, Sentinel-2, Salesforce CRM, XGBoost yield models) and deliver measurable business outcomes. 2. Engagement Structure Type
Remote contractor / part-time to full-time based on performance
Location
Ukraine (remote) — overlap with U.S. Eastern Time required (min. 4 hrs/day)
Start
45 days
Duration
Initial 6-month contract with renewal; potential for long-term engagement
Compensation
Competitive rate, milestone-based; commensurate with seniority 3. Primary Responsibilities3.1 Agent Development * Design, build, and deploy multi-agent workflows using LangGraph, CrewAI, and Anthropic MCP * Implement modular, decoupled agent architectures with clearly defined handoff points (human-in-the-loop) * Wrap legacy data APIs into agent-accessible interfaces without disrupting existing production stacks * Build data ingestion pipelines (Prefect/Airflow) that feed real-time context into agentic workflows * Integrate agents with Salesforce CRM, Sentinel-2 satellite imagery, ERA5 weather data, and XGBoost yield models
3.2 Multi-Agent Orchestration * Orchestrate specialized micro-agents across 11 defined agent use cases * Define inter-agent communication protocols, state management, and error-recovery logic * Implement agent observability: logging, tracing, cost tracking, and performance benchmarking
3.3 LLM & Prompt Engineering * Design and iterate production-grade prompts for classification, extraction, summarization, and generation tasks * Apply RAG (Retrieval-Augmented Generation) patterns against CERES AI’s internal knowledge bases and agronomic documentation * Evaluate and fine-tune models (GPT-4o, Claude, Gemini) for specific agricultural domain tasks
3.4 DevOps & Deployment * Containerize agents using Docker and deploy on cloud platforms (AWS, GCP, or Azure) * Implement CI/CD pipelines for agent code; enforce version control best practices (Git) * Monitor deployed agents with alerting; maintain SLAs for production workflows
4. Required Technical Skills The following matrix reflects the core competencies evaluated during the hiring process:
LLM Frameworks
LangGraph (stateful workflows), CrewAI (role-based agents), Anthropic MCP (tool/data connections) — ALL THREE required
Orchestration
Prefect or Apache Airflow for pipeline scheduling and data workflow integration
Languages
Python (primary — 4+ years production), JavaScript/TypeScript (secondary)
RAG & Vector DBs
LlamaIndex or LangChain for RAG; Pinecone, Weaviate, or pgvector for vector storage
ML / AI Models
Experience with XGBoost, scikit-learn, fine-tuning LLMs; familiarity with geospatial/satellite ML a strong plus
APIs & Integration
REST API design, Salesforce API integration, webhook handling, event-driven architecture
Cloud & DevOps
AWS (Lambda, S3, EC2) or GCP; Docker; CI/CD (GitHub Actions); infrastructure-as-code a plus
Data Engineering
Pandas, NumPy, SQL; experience with time-series data (weather, satellite NDVI) strongly preferred
NLP
Named entity recognition, intent classification, document parsing, transcript analysis
5. Preferred Background & Experience5.1 Must-Have * 3+ years building and shipping production AI/ML systems (not just notebooks or demos) * Demonstrated experience with at least one major LLM orchestration framework (LangGraph, CrewAI, or AutoGen) * Strong Python knowledge and engineering — clean, documented, testable code * Experience integrating LLM agents with external data sources (APIs, databases, document stores) * Proficiency in English (written and spoken) — will participate in U.S. team standups and documentation
5.2 Strongly Preferred * Experience in AgTech, InsurTech, FinTech, or any domain with structured geospatial or time-series data * Familiarity with satellite imagery pipelines (Sentinel-1/2, NDVI, SAR) or weather data (ERA5, CHIRPS) * Prior work with Salesforce APIs or CRM-integrated AI workflows * Contributions to open-source AI/ML projects or published technical writing * Experience with parametric insurance logic, claims automation, or agricultural risk modeling
5.3 Educational Background * Bachelor’s or Master’s degree in Computer Science, Data Science, Software Engineering, or equivalent * Equivalent demonstrated experience (strong portfolio / GitHub) accepted in lieu of formal degree
6. Working Norms & Collaboration * Weekly 1:1 with PM (async-friendly; async standups) * Work tracked via Linear or Jira; all code committed to private GitHub organization * Daily async check-in via Slack with a 4-hour Eastern Time overlap window * Documentation-first culture — all agent designs require a brief technical spec before implementation * Each agent must ship with: unit tests, an integration test, and an observability hook (logging/tracing) * Milestone-based deliverables tied to agreed sprint cycles (2-week sprints)
