Computational Designer — New York, NY
Зарплата не указана
ОткликнутьсяКомпания: ••••••••
Локация: New York, NY
Формат: не указан
Занятость: Полная занятость
Грейд: не определён
PaVrd
PaVrd is a nature-based research and design company based in Manhattan. We incubate ventures and technologies that reimagine the relationship between humanity and the natural world. Working across disciplines—from architecture and ecology to computation and materials science—we develop nature-centric solutions to critical environmental challenges.
EDEN
Existing environmental tools for site planning and development sit on either side of a gap: remote sensing observes what is already on the ground, and traditional simulations evaluate fixed plans someone has already drawn. EDEN is building a generative ecosystem digital twin that works from both the existing site and target outcomes, such as carbon sequestration, stormwater retention, microclimate cooling, biodiversity and the cost of site development, to discover the optimal earthworks, planting and management sequence that achieves them.
Our methods already work on real sites. Over the past several years we have embedded with developers, landowners and operators, designing and engineering their sites in closer harmony with the ecosystems around them, and building the tools we needed along the way. Those tools now generate road networks and lot layouts, minimize earthworks, classify the vegetation a site can support, simulate how an ecosystem performs over multi-year horizons, and optimize a design against cost and ecological outcomes. AI agents adapt these design processes to each client’s brief, and numerical optimizers search them for the designs that perform best.
Each of these design processes is a model: a process that turns a site and a set of decisions into a design and measures how it performs. We already have working graphs for residential development, reforestation, mine reclamation, wildlife crossings, wildfire protection and flood risk. Every new client problem needs a graph built or adapted for it, and we are seeking a Computational Designer to build them.
Role Overview
PaVrd is seeking a Computational Designer to join the EDEN team and turn each client’s problem into a model that EDEN can solve. The designer builds a procedural graph, and the agents and optimizers search it.
A model has four parts, and the designer owns all four:
• The site: terrain, soils, water, existing vegetation, roads and boundaries, built from geospatial data.
• The design moves: the interventions on the site, such as regrading, planting, buildings, road and path networks, or drainage. Each move is a slot that an agent fills with a program, or a set of parameters an optimizer tunes.
• The evaluation: the simulations and measures that score a design, such as runoff, habitat connectivity, fire spread, biomass or cost. Some are rules a design must meet; others are objectives to improve.
• The adaptation surface: what a client’s brief is allowed to change. This covers which numbers it may move and within what bounds, which rules it may tighten or switch off, and what guidance the agents receive.
This role sits at the intersection of computational design, ecology, the built environment, and applied AI. The designer will work closely with clients to define the problem, with EDEN’s research and development teams to make the measures right, and with the System Architect and software engineers to make graphs reusable from one project to the next. The role is hands-on: the designer builds and runs the models themselves, from site data to the design a client acts on.
This role is based on-site in the EDEN office in New York City.
Key Responsibilities
• Frame client problems: Work with clients, architects, engineers, ecologists and planners to turn a project’s goals into a searchable problem: what can change on the site, what must hold, what to improve, and what it may cost.
• Build procedural graphs: Author graphs that go from site data to a scored design, combining procedural geometry, simulation and data analysis. Make them fast and robust enough to be run thousands of times by an automated search.
• Design for search, not only for one answer: Expose the design moves as program slots and parameters, and give each a sensible range. Write clear guidance for the agents that fill those slots. Decide what a client’s brief may change.
• Develop and validate the evaluation: Build and calibrate the measures that score a design, together with our ecologists and science team. Check that the measures and a design’s numbers agree with each other and with field data where we have it. Close the loopholes that would let a search score well without solving the problem.
• Run and read searches: Run agentic and numerical searches on your graphs and read their records: which designs were tried, why the best one won, and where the search got stuck. Feed that back into the graph.
• Communicate results: Turn a search’s results into designs, maps and trade-offs that clients can act on. Present them to technical and non-technical audiences.
• Generalize the methods: Turn one-off project graphs into reusable graphs and components that other designers and future projects can adapt. Give the platform team concrete feedback on the tools.
Key Goals & Outcomes
• First 90 Days:
• Learn our existing graphs end to end, from site data to delivered design. Run searches on them and read the results.
• Adapt an existing graph to a new site or client brief, and take it through a search to a result we can present.
• Build your first new graph for a live client problem, together with an ecologist and a platform engineer.
• Document what made it hard, and propose one improvement to how we build, check or reuse graphs.
Required Experience
• An advanced degree or equivalent professional experience in architecture, landscape architecture, computational design or a related field.
• 3 to 6 years of experience in computational design and simulation or a related field, with a portfolio of procedural or generative design work applied to sites, landscapes or the built environment.
Preferred Experience
• Houdini: procedural modelling with SOPs, VEX and Python, digital assets and PDG. Our graphs are built in Houdini, and we will help you learn it.
• Experience with optimization or search in design, such as Galapagos, Wallacei, Octopus, or Bayesian, evolutionary or multi-objective methods.
• Working with LLMs or AI agents in a design workflow.
• Domain simulation in a design context: hydrology and flood, microclimate, fire, ecological succession or habitat connectivity.
• Rhino and Grasshopper.
• Grading, earthworks or planting design on real projects.
• A background or strong interest in ecology or environmental science.
Technical Skills
• Strong procedural or computational design skills, and fluent scripting in Python.
• Working knowledge of computational simulation methods, such as numerical solvers for flow, diffusion or heat, agent-based or cellular models, and optimization. You understand their assumptions and where they stop holding.
• Experience with GIS and data analysis workflows: preparing raster and vector data, terrain and hydrology analysis, and analysing results with tools such as QGIS, ArcGIS, GDAL or Python’s data libraries.
Essential Qualities
• The habit of checking whether a model’s numbers are correct, not only whether the geometry looks right.
• Clear communication with clients, scientists and engineers, in writing and in presentations.
Why this role is unusual
In most practices, a computational designer builds a script that produces a design. Here, you build the problem itself, and AI agents and optimizers search it, often testing hundreds of designs a person would never draw.
That changes what good work looks like: a graph has to be fast and robust under automated search, its measures have to hold up in the field and its results have to stay legible to the client who acts on them.
You will help define what computational design means for ecological engineering, on projects whose outcomes are measured in water retained, habitat connected and carbon stored.
Compensation Range: $118K - $147K