Lead AI Engineer (Member of Technical Staff)

US startup, global team. Highly senior individual contributor; may lead to a management role.

About Plankore

Plankore is a VC-funded US-based startup building AI that truly understands construction blueprints — not just pixels, but structure, intent, and cross-sheet reasoning.

Construction blueprints are the system of record for a $13T industry, yet today's tools still rely on manual tracing or shallow automation. We're building a multimodal reasoning layer that can read plans, legends, notes, and sections to produce trustworthy quantity takeoffs for contractors and suppliers.

Plankore is a daily-use system for contractors making high-stakes decisions. Accuracy, usability, and trust matter.

This role owns the intelligence engine at the core of the product.

The role

As a Lead AI Engineer (Member of Technical Staff), you will own and drive the core intelligence engine behind Plankore. Construction drawings present complex spatial, visual, and text alignment challenges; you will design, train, and deploy multimodal models capable of reasoning over complex 2D spatial layouts, multi-page document structures, and dense domain-specific text.

You will work directly with:

• US-based CEO (15+ YOE in Product/Eng, Chicago Booth MBA, NUS B.Eng)
• Dual US/Vietnam-based CTO (15+ YOE in Computer Vision/Eng, CMU PhD)

and other technical members of the company. This is a high-ownership role with independent execution expectations and a direct trajectory toward engineering management as the team grows.

What you'll own

• Multimodal core architecture: lead the technical strategy and R&D for custom computer vision, vision-language models (VLMs), and LLM architectures tailored to blueprint understanding, visual grounding, and spatial reasoning.
• 0-to-1 model development: drive the entire R&D pipeline — from synthetic/real data curation and preprocessing to pre-training, fine-tuning (SFT/RLHF/DPO), and parameter-efficient adaptation (LoRA/QLoRA).
• Cross-modal alignment: develop novel techniques to align high-resolution visual engineering diagrams with domain-specific text, tables, legends, and section callouts.
• Production deployment & inference: optimize state-of-the-art vision and language models for production-scale, low-latency execution (quantization, distillation, model pruning, FSDP/DeepSpeed).
• Autonomous execution: independently identify technical bottlenecks, conduct rapid experiments, and ship robust ML models from research to high-stakes production.

What we're looking for

• Experience / education: Ph.D. in computer science, machine learning, computer vision, or a related technical field, or 10+ years of hands-on professional experience building and deploying machine learning systems at scale.
• Computer vision (CV): object detection, segmentation, vision transformers (ViT), and high-resolution document/diagram analysis.
• Vision-language models (VLMs): multimodal pre-training, image-text alignment, visual grounding, and cross-modal reasoning (e.g., CLIP, LLaVA, BLIP).
• Large language models (LLMs): transformer architectures, prompt engineering, structured output generation, and post-training fine-tuning.
• Engineering proficiency: deep expert-level proficiency in PyTorch, distributed training frameworks (DeepSpeed, Megatron-LM, FSDP), and modern GPU acceleration pipelines.
• Startup mindset: high agency, strong bias for action, zero reliance on step-by-step supervision, and comfort with ambiguity.

You should be able to independently lead and execute efforts across all three of CV, VLMs, and LLMs.

Nice to have

• Experience with technical document processing (PDFs, CAD files, engineering blueprints, high-resolution schematics)
• First-author publications in top-tier conferences (CVPR, NeurIPS, ICML, ICCV, ECCV, ACL)
• Experience with synthetic data generation pipelines for training visual models

Working model (important)

• This is a Vietnam-based role working closely with a US founding team.
• You'll collaborate directly with experienced US product and engineering leaders.
• Take ownership of your work.
• Clear communication (Vietnamese or English) and async collaboration are essential.

Why this role is exciting

• High autonomy: direct influence on company direction and core AI intellectual property, with zero legacy constraints.
• Massive industry impact: build real-world AI that powers critical workflows in a $13T global industry.

Compensation

• Engagement: full-time (preferred), or part-time
• Competitive compensation for Vietnam senior/lead talent

How to apply

Send your resume to hello@plankore.ai with the role title in the subject line.