Head of Computer Vision

San Franciscofull-timeexecutive

OtherComputer Vision

0 views0 saves0 applied

Apply Now

Quick Summary

Overview

Position Overview Bobyard solves the hardest CV problems in construction. Contractors burn dozens of hours per project doing takeoffs by hand. We automate it with model systems that have to be better than human — on real, ugly, million-dollar drawings.

Key Responsibilities

Build SOTA CV systems: Architect models that detect, segment, OCR, and reason over construction drawings. One error costs millions. Own end-to-end pipelines: Data ingestion, labeling, training, eval, and low-latency inference.

Technical Tools

segment

Bobyard solves the hardest CV problems in construction. Contractors burn dozens of hours per project doing takeoffs by hand. We automate it with model systems that have to be better than human — on real, ugly, million-dollar drawings.

We’ve built a high-performing CV team attacking the most non-trivial problems in applied vision. Now we need a leader to set technical direction, enforce a ruthless focus on impact, and 10x our execution speed and quality.

This is not a “manage from afar” role. The bar is: you code as well as our strongest ICs. You’ll lead strategy, grow the team, and keep a monster hiring bar. You’re still in the weeds — because that’s where leverage lives.

Responsibilities

~1 min read

→
Build SOTA CV systems: Architect models that detect, segment, OCR, and reason over construction drawings. One error costs millions.
→
Own end-to-end pipelines: Data ingestion, labeling, training, eval, and low-latency inference. If it doesn’t run in prod at scale, it doesn’t exist.
→
Scale the infra: Train on massive datasets and serve millions of inferences. Drive cost down while accuracy goes up. Make the hard systems calls.
→
Guarantee production reliability: Uptime, latency, drift. When models break, revenue stops. You’re accountable for every prediction.
→
Push model performance: Accuracy, robustness, generalization. 10,000+ drawing formats, scanned, crooked, handwritten — they all need to just work.
→
Ship with product + fullstack: No research silos. You integrate models into features customers use daily. Tie every experiment to user impact.
→
Attack failure modes: Live in real customer data. Find where we fail, why we fail, and fix it this week.
→
Apply research pragmatically: Stay on the frontier. Rip what works from papers, ship it next sprint, measure impact in dollars saved.
→
Set technical strategy: Define the ML roadmap. Decide what we build, what we buy, what we kill. Align the team on the highest-leverage bets.
→
Lead and grow the team: Hire A+ ML talent. Coach, unblock, and develop world-class engineers. Give hard feedback. Keep the bar savage.

Founder-level ownership: You treat model performance like it’s your P&L. No one tells you what to fix — you already shipped the fix.
Deep CV + DL authority: You’ve built and scaled vision systems in production. Segmentation, detection, document AI, VLMs — you’ve done it for real.
Production-obsessed: You’ve put ML into products with real SLAs. You know the last 5% costs 10x and you know how to get it anyway.
Messy data killer: Real drawings are garbage. You love it. You’ve turned chaos into datasets that train models that print money.
Research-minded, execution-obsessed: You read papers on Sunday and ship code on Monday. Impact > citations.
Proven technical leader: You’ve led a world-class CV/ML team before. People follow you because you’re right and you ship.
Startup work ethic: This isn’t a cushy big-tech director role. We’re at war with manual takeoffs in a $13T industry. You want that.
Learning machine: New architecture drops, you’re testing it same day. You navigate unknown spaces and make them legible for the team.

This is full-time & in-person in SF. Learning rate and dedication beat pedigree. If you can prove you’ll lead us to ship models customers trust with million-dollar bids — at the speed and quality the market demands — we want you.