Muro AI
About Muro AI
Muro AI is transforming how the $2T construction industry plans and builds. Founded by Cornell alumni, ex-founders, and former McKinsey operators, we’re building AI agents that automate the most complex, manual, and costly phase of construction: preconstruction.
We move fast, build with conviction, and obsess over delivering real impact to the people who build our world. If you want to define how AI understands and reasons over architectural and design documents at scale, this is where it starts.
Why This Role Exists Crack the hardest problems in construction tech: enabling AI to deeply understand architectural documents—and then act on them. We’re building systems that don’t just read plans, but reason across them and execute complex workflows using specialized industry tools.
Shape a category:
No one owns “pre-con AI” yet. You’ll build the core research—from document understanding to agentic execution—that sets the foundation.
Move the industry:
Your work will help contractors win multimillion-dollar bids, eliminate scope gaps, and unlock massive productivity gains.
Who You Are
A research-minded builder with 4–6 years of deep technical ML experience.
You’re comfortable in the messy 0→1 phase: half-baked specs, ambiguous problem statements, and solving foundational technical challenges that will define the next decade of construction AI energize you.
You’re curious, rigorous, and want to own hard problems end-to-end—from research to production. You communicate complex work clearly to engineers, product, and leadership; you’re the go‑to scientist the team turns to when they need to understand what’s possible.
Your Mission 1. Build the document understanding layer that powers everything downstream
Create the foundational multimodal system that all downstream agents depend on. Research and build models that understand complex construction‑specific documents formats – using custom fine tuned multimodal models and RAG architectures. Architect pipelines for OCR, image segmentation, layout analysis, entity resolution… Build rigorous evals and benchmarks from real‑world documents.
2. Push agentic AI into production
Context engineer construction the agent harnesses for multi‑step construction workflows. As well as, architect and expose novel tools. Own the full path from prototype to deployment alongside the AI engineering team.
3. Make AI reliable for million‑dollar decisions
Build labeled datasets with construction SMEs. Develop hallucination‑detection, confidence calibration, and evaluation frameworks tailored to the most complex AEC tasks.
Who Should Apply We want research‑driven builders who love pairing deep model work with real customer understanding.
You might be:
A
Research Engineer/Scientist
looking to see your work deployed, not just published
An
ML Engineer
or
Applied Scientist
who wants to push the frontier on multimodal and agentic systems
A
Software Engineer
who has transitioned into deep AI/ML research and wants to own problems end-to-end. And wants to productionize their new skill set.
A
founding‑team caliber engineer
eager to tackle the hardest technical challenges in AEC tech
You’ll Stand Out If You Have
Prior 0→1 or founding‑team experience at a startup
Experience working directly with users or in forward‑deployed environments
Built a multimodal model or document‑understanding pipeline used in production
Hands‑on experience with RLHF, DPO, SFT, or parameter‑efficient fine‑tuning methods
Advanced OCR/CV experience (layout‑aware models, document transformers, segmentation)
Designed or scaled RAG systems for complex, multi‑document retrieval
Experience with agentic LLM systems, tool use, or multi‑step reasoning harnesses
Background in architectural/engineering drawings (AEC experience is a plus, not required)
Location:
In‑person in San Francisco. Relocation provided.
Compensation:
$175K–$250K base meaningful equity
#J-18808-Ljbffr
We move fast, build with conviction, and obsess over delivering real impact to the people who build our world. If you want to define how AI understands and reasons over architectural and design documents at scale, this is where it starts.
Why This Role Exists Crack the hardest problems in construction tech: enabling AI to deeply understand architectural documents—and then act on them. We’re building systems that don’t just read plans, but reason across them and execute complex workflows using specialized industry tools.
Shape a category:
No one owns “pre-con AI” yet. You’ll build the core research—from document understanding to agentic execution—that sets the foundation.
Move the industry:
Your work will help contractors win multimillion-dollar bids, eliminate scope gaps, and unlock massive productivity gains.
Who You Are
A research-minded builder with 4–6 years of deep technical ML experience.
You’re comfortable in the messy 0→1 phase: half-baked specs, ambiguous problem statements, and solving foundational technical challenges that will define the next decade of construction AI energize you.
You’re curious, rigorous, and want to own hard problems end-to-end—from research to production. You communicate complex work clearly to engineers, product, and leadership; you’re the go‑to scientist the team turns to when they need to understand what’s possible.
Your Mission 1. Build the document understanding layer that powers everything downstream
Create the foundational multimodal system that all downstream agents depend on. Research and build models that understand complex construction‑specific documents formats – using custom fine tuned multimodal models and RAG architectures. Architect pipelines for OCR, image segmentation, layout analysis, entity resolution… Build rigorous evals and benchmarks from real‑world documents.
2. Push agentic AI into production
Context engineer construction the agent harnesses for multi‑step construction workflows. As well as, architect and expose novel tools. Own the full path from prototype to deployment alongside the AI engineering team.
3. Make AI reliable for million‑dollar decisions
Build labeled datasets with construction SMEs. Develop hallucination‑detection, confidence calibration, and evaluation frameworks tailored to the most complex AEC tasks.
Who Should Apply We want research‑driven builders who love pairing deep model work with real customer understanding.
You might be:
A
Research Engineer/Scientist
looking to see your work deployed, not just published
An
ML Engineer
or
Applied Scientist
who wants to push the frontier on multimodal and agentic systems
A
Software Engineer
who has transitioned into deep AI/ML research and wants to own problems end-to-end. And wants to productionize their new skill set.
A
founding‑team caliber engineer
eager to tackle the hardest technical challenges in AEC tech
You’ll Stand Out If You Have
Prior 0→1 or founding‑team experience at a startup
Experience working directly with users or in forward‑deployed environments
Built a multimodal model or document‑understanding pipeline used in production
Hands‑on experience with RLHF, DPO, SFT, or parameter‑efficient fine‑tuning methods
Advanced OCR/CV experience (layout‑aware models, document transformers, segmentation)
Designed or scaled RAG systems for complex, multi‑document retrieval
Experience with agentic LLM systems, tool use, or multi‑step reasoning harnesses
Background in architectural/engineering drawings (AEC experience is a plus, not required)
Location:
In‑person in San Francisco. Relocation provided.
Compensation:
$175K–$250K base meaningful equity
#J-18808-Ljbffr