ML Engineer, Agentic Systems
On-site ML Engineer in Houston at Viridien (formerly CGG) building production agentic LLM applications that automate geoscience workflows: designing multi-agent workflows, tool/API integration, persistent state and human-in-the-loop controls, plus LLM fine-tuning (LoRA, DPO, GRPO), delivered as services with FastAPI, PostgreSQL and Docker.
- Compensation
- Not disclosed
- Location
- Houston, United States
- Work setup
- Onsite
- Seniority
- Not specified
- Employment
- Not specified
- Experience
- Not specified
- Category
- Ml / AI
- Posted
Skills
Must have
- Bachelor's degree or higher in Computer Science, Engineering, Physics, Mathematics, Geophysics, or related field
- Minimum cumulative GPA of 3.5/4.0 for all earned degrees
- Strong programming and software engineering fundamentals, including asynchronous and concurrent programming
- Sound design and product judgment: explain why a system is built the way it is and how its quality is measured
- Hands-on experience building multi-agent or graph-based LLM workflows
- Experience designing how agents find information: tool-based search over files, logs, databases, semantic retrieval with reranking
- Practical LLM fine-tuning experience: parameter-efficient methods (LoRA), dataset preparation, offline/online evals, working understanding of RL-based fine-tuning
- Experience with tracing and evaluation of LLM applications
- Experience building web services and APIs (e.g., FastAPI), working with PostgreSQL, Docker, and Git
- Ability to harness AI coding agents such as Claude Code or Codex: guide execution, review output, own resulting code
Nice to have
- Experience designing complex agent systems: task decomposition, parallel execution, result aggregation, human-in-the-loop orchestration
- Hands-on LLM post-training experience (e.g., DPO, GRPO) including reward design for verifiable outcomes and multi-step tool use
- Experience running enterprise-scale applications in production: retries, idempotency, timeouts, rate limits
- Prior work in workflow automation; exposure to geoscience or scientific computing is a plus
- Frontend experience building lightweight operator dashboards
Full description
Viridien () is an advanced technology, digital and Earth data company that pushes the boundaries of science for a more prosperous and sustainable future. With our ingenuity, drive and deep curiosity we discover new insights, innovations, and solutions that efficiently and responsibly resolve complex natural resource, digital, energy transition and infrastructure challenges.
Job Summary
We are hiring an ML Engineer to build production agentic applications that automate complex geoscience workflows. You will design agent-based systems, integrate tools and external APIs, implement persistent state management, provide agents with the right context, and deliver well-tested services with lightweight web interfaces for operators and end users.
Beyond implementation, you will help shape what these applications do: which workflows to automate, where people stay in the loop, and how results are presented and verified. You will also be responsible for designing and executing LLM fine-tuning strategies when model adaptation is required.
About the Team
We work together to Build high-impact tools and platforms that drive innovation – from scientific seismic algorithms to data visualization tools to intuitive web apps. combining our expertise across regions to drive innovation and impactful results. Our team values collaboration, continuous learning, and innovation, and we take pride in supporting each other to achieve shared goals.
Key Responsibilities
- Shape the application logic and user experience: define what a good result looks like, remove what does not help users decide, and make outputs easy to verify.
- Design agent workflows with branches, loops, sub-workflows, and explicit routing, choosing the simplest structure that solves the problem.
- Implement human-in-the-loop approvals, and decide which actions an agent may take on its own.
- Give agents the right context: tool-based search over files, logs, code, and databases, using text search, SQL, and semantic retrieval where appropriate.
- Ground answers in cited evidence, enforce tool permissions, and guard against prompt injection and unsafe actions.
- Integrate tools and external APIs. Handle concurrency, retries, and error paths.
- Add database-backed persistence. Ensure workflows can pause and resume reliably.
- Design and execute LLM fine-tuning strategies when needed: data curation, supervised fine-tuning with parameter-efficient methods such as LoRA, preference optimization (e.g., DPO), reinforcement learning methods such as GRPO with verifiable rewards, evaluation sets, and rollback criteria.
- Instrument agent runs with tracing and continuous evaluation, making decisions and actions auditable and detecting quality regressions early.
- Deliver production-ready services with automated tests and CI/CD, and expose APIs where other teams or systems need them.