Machine Learning Engineer

Conduit; San Francisco (CA); On-site; USD 150,000 - 210,000; Full time

Benefits offered by this job

Top pay

Equity

Job summary

Conduit is building autonomous factories through a unified data layer, delivering AI across machines and software. We seek a customer-first CV/ML engineer to own real-time streaming systems that classify and act on factory signals, from data collection to edge inference deployment.

You will lead end-to-end pipelines, from labeling to training to evaluation, and coordinate hardware orders for edge compute. Strong Python and PyTorch expertise are essential, with a focus on industrial applications.

Qualifications

• 4+ years of applied ML / CV work, including production deployments.
• You have shipped at least one inference pipeline that someone else depended on.
• Strong Python and PyTorch; familiar with CV concepts and inference toolchains.

Responsibilities

• Own customer CV/ML deployments end to end. Scope the problem, install cameras and edge compute, build the data pipeline, train the model, deploy inference, validate on the floor.
• Build end-to-end pipelines from training to inference. Data collection, labeling (manual and automated), training, evaluation, deployment, monitoring.
• Stream and process video at industrial volume. Multi-camera setups, real-time inference on the edge, smart triggers, alerting.
• Apply CV to industrial problems: defect detection, classification, action/ergonomics analysis, machine state, material tracking, operator-assist.
• Spec and order the hardware to run it. GPUs, edge boxes, cameras, lenses, lighting.
• Translate customer needs into clean engineering briefs for the SF team. Build relationships with plant managers and operators so the system gets used.

Skills

Python

PyTorch

CV/ML pipelines

Edge inference

Tools

TensorRT

ONNX

Triton

Vision transformers

Job description

Conduit builds autonomous factories through a single data abstraction layer across every machine, robot, sensor and software a factory might have, agnostic of brand or backend.We partner with manufacturers to turn their existing infrastructure into AI without rip and replace, and are able to light up an entire factory in a matter of days, not months, with as few as 2 FDEs at a time. We are live with manufacturers across the country and are growing quickly. We are hiring the people who are excited to tackle challenges at the intersection of data science, industrial automation, and software engineering and bring AI to American manufacturing.

The Role

This is a customer-first role applied to computer vision and ML. Live factory data streams, including defects, ergonomics, cycle times, machine state, operator behavior, and material flow, will drive your research, and you will stand up the streaming vision system that captures, classifies, and acts on those signals in real time. You are technical enough to own the full pipeline: data collection on real factory hardware, manual and automated labeling, model training and evaluation in PyTorch, inference deployment on edge hardware you specced and ordered.

What you'll do

• Own customer CV/ML deployments end to end. Scope the problem, install cameras and edge compute, build the data pipeline, train the model, deploy inference, validate on the floor.
• Build end-to-end pipelines from training to inference. Data collection, labeling (manual and automated), training, evaluation, deployment, monitoring.
• Stream and process video at industrial volume. Multi-camera setups, real-time inference on the edge, intelligent downsampling, smart triggers, alerting.
• Apply CV to industrial problems: defect detection, classification, action / ergonomics analysis, machine state, material tracking, operator-assist.
• Spec and order the hardware to run it. GPUs, edge boxes, cameras, lenses, lighting. You are comfortable being trusted with a $40K compute order.
• Translate customer needs into clean engineering briefs for the SF team. Build relationships with plant managers and operators so the system gets used.

Who You Are

• 4+ years of applied ML / CV work, including production deployments. You have shipped at least one inference pipeline that someone other than you depended on.
• You have built end-to-end pipelines from training to inference, including the data labeling muscle - both manual workflows and automated / programmatic labeling, weak supervision, pseudo-labeling, active learning.
• Strong Python. Strong PyTorch. You are conversational with modern CV, detection, segmentation, classification, action recognition, vision transformers, multi-modal, and the relevant inference toolchain (TensorRT, ONNX, Triton, etc.).
• You are comfortable around hardware and are confident prescribing hardware across a diverse breadth of cotnexts. You have selected cameras, ordered GPUs, set up edge inference boxes.
• You are great with customers. You can do a plant tour, push back on a CEO, and earn the trust of an operator who has never seen a model before.
• You want to be a part of the reindustrialization of the United States

Bonus Points

• Experience with industrial machine vision (Cognex, Keyence, Zebra / Adaptive Vision, MVTec Halcon).
• Experience with NVIDIA Jetson, Orin, Isaac, Triton, TensorRT, DeepStream.
• Experience streaming video at scale: RTSP, GStreamer, multi-cam sync, edge-to-cloud pipelines.
• Experience with ergonomics, pose estimation, action recognition, or video understanding.
• Experience with annotation platforms (Scale, Labelbox, V7, internal tools) and building automated labeling loops.

Why Conduit?

This is the rare opportunity to work at the forefront of Physical AI at a company with traction and a real technical moat. At Conduit, you will train your model and see its benefits in the real world.

• Top tier pay and equity: We want to make it easy to join Conduit.
• Generational opportunity. Be early at a company defining the software layer for industrial automation.
• American re-industrialization is happening. We're building the platform to make it real - faster.
• Zero bureaucracy. High impact. You'll be on the front lines, solving real problems for real factories.
• You will level up. You'll work directly with the founder and exec team and grow into a senior operator

Conduit is an equal opportunity employer. We welcome applications from all qualified candidates.