Founding AI Engineer

About the role

We are seeking a Founding AI Engineer to join our team in San Francisco. This role focuses on applied AI/ML engineering, specifically within computer vision and multimodal systems. The ideal candidate has shipped agentic multimodal systems to real users and is comfortable building production Vision-Language Model (VLM) pipelines with real hardware constraints.

Key Responsibilities

• Build and ship production agentic-VLM pipelines running on industrial smart glasses, including multi-step, tool-using visual-reasoning loops against real customer workflows (SOPs, inspection, field service).
• Own model orchestration and runtime optimization for edge inference, balancing model quality against latency with graceful fallback across connectivity conditions.
• Design and build the evaluation harness and data flywheel from scratch, including failure-mode capture, customer-data fine-tune loops, and measurable model quality improvements.
• Ship real-time voice-video AI interfaces adapted to different end-user profiles (video-heavy, conversational speech, and proactive alerts).
• Build RAG pipelines for the efficient creation and querying of enterprise knowledge bases from field operator data.
• Drive multimodal model training for on-premise deployments, including open-source model SFT, RL post-training, and quantization.

Requirements

• Experience: 3-5 years of experience in applied AI/ML engineering, ideally in computer vision or multimodal systems.
• Production Track Record: Demonstrated experience shipping multimodal and computer vision systems in the VLM era (production, not demos or pure research). Must have owned the model layer end-to-end.
• Background: Experience at a startup or AI team building production AI products is preferred. Big-company tenure (e.g., Meta Reality Labs, Snap, Apple Vision, Google) is a bonus only if on a directly relevant team (AR/smart-glasses, real-time video/streaming, on-device/edge ML) or paired with a builder signal (founder/early-startup/side projects/OSS).
• Domain Experience: Production AR/wearable AI experience or autonomous driving Computer Vision is preferred. Industrial domain exposure (data centers, energy grid, aerospace, manufacturing) is a plus.
• Technical Skills: Applied VLM/Multimodal Engineering: Shipping, hardening, and applied fine-tuning of vision-language/video-language models. Applied Agentic AI/Model Orchestration. Rigorous evaluation disciplines (ground-truth, trajectory/tool-call accuracy, regression). On-prem/self-hosted model deployment and optimization. In-context grounding/RAG against knowledge bases.
• Education: Strong CS/ML/Engineering background or demonstrated equivalent shipping record. Master’s with a vision or multimodal research component is preferred.
• Technical Stack: Python, PyTorch, TensorFlowvLLM, Triton, Ray Serve, ONNX, TensorRT, Hugging Face Transformers, LangChain, RAG, RLHF, SFT, Quantization (GPTQ, AWQ), Edge AI, Multimodal LLMs, Vision-Language Models, Docker.

Technical Stack

• Python, PyTorch, TensorFlowvLLM, Triton, Ray Serve, ONNX, TensorRT
• Hugging Face Transformers, LangChain, RAG, RLHF, SFT, Quantization (GPTQ, AWQ)
• Edge AI, Multimodal LLMs, Vision-Language Models, Docker

Additional Information

• Visa Sponsorship: H-1B transfers and TN visas are supported. No new H-1B sponsorship.
• Location Requirement: Candidates must currently reside in the USA or Canada.
• Work Style: Ideally willing to join a hacker house (live on site). At minimum, must be willing to work on-site 5 days per week in San Francisco.
• Referral Bonus: $5,000 for successful placements.

How to Apply

If you have the required skills and experience, please send your resume with details to: 📧 skumar@cognistack.co