About the roleML sits at the heart of Treeswift's product translating complex sensor data into meaningful insights that drive critical decision-making for our customers. As our business grows the importance of this work only grows with it.
You will join at a critical inflection point and will lead the charge on the following priorities (amongst others):
- Technical Leadership. You will be responsible for continuing to improve the quality and consistency of our models across a growing breadth of operating conditions. As the breadth of models and datasets grows you will thoughtfully introduce feedback loops and ML Ops that accelerate model improvement. You have excellent judgement on what problems require fast experimentation/validation, and where we should be focused on sustainable scalable improvement over a longer period of time.
- Edge model development. Working closely with our robotics and embedded teams, you will support growing exploration of model deployment on devices for real-time use cases.
- Lead and grow our team (as needed). We have an incredibly talented ML team distributed across our NYC, Philly and SF offices. You will lead the current team, identify gaps and grow the team as needed.
- Collaborate closely with our Software, Robotics, Hardware and Product teams to deliver on Treeswift's vision and business objectives.
You will report to our Head of Software Engineering. This is a full-time, hybrid role based out of our Lower Manhattan, NYC office (2-3 days per week in person, currently pinned to Tuesdays and Wednesdays).
About Machine Learning At TreeswiftMachine Learning at Treeswift is the layer that translates sensor data into meaning for our end-users. At a high-level we develop three categories of model that power downstream business decisions.
- Classifying a point cloud - distinguishing ground from vegetation from structures etc.
- Identifying individual objects within 3D space - poles, equipment, wires, trees, etc.
- Describing characteristics of individual objects. This includes everything from tree species and health, detailed wire types, pole information (height, attachments, etc.) etc.
Together this information informs critical decision-making for our customers such as:
- What vegetation is a risk to utility infrastructure and how severe is that risk? What action is recommended as a result (trim, removal, etc...)?
- Is ongoing asset construction deviating from the expected design and if so how? If so, is this an approved deviation or a quality issue that needs to be addressed before a downstream customer or financial impact?
Required skills and experience- Early-stage or fast-moving environments where scope and ownership evolve quickly.
- Strong familiarity with models built on images, video, lidar, pointclouds, speech, or geospatial data sources.
- Building and using cloud-based training and inference pipelines (including orchestration, deployment, and the operational realities of running models at scale).
- Willingness to drive architecture and approach across both, bring innovative ideas and rapid execution to new and existing models, and be accountable for the quality and consistency of our models across a growing breadth of operating conditions.
- Leading development of our training and inference pipelines. You will lead how we build and evolve these pipelines as well as move us towards on-device inference for time-sensitive use cases.
- 3+ years experience in leadership: setting clear direction for a team, driving architecture and design decisions, reviewing others' work, and mentoring individual engineers.
- Experience creating, curating, and cleaning training datasets, and building the evaluation and feedback loops that drive model improvement.
- Strong programming skills and expertise in deep learning frameworks such as PyTorch, scikit-learn, and TensorFlow.
- Excellent communication and collaboration skills. You have the ability to keep a team distributed across three offices (NY, Philly, SF) well-aligned. You will be a strong partner to the product team on how to balance technical trade-offs and product objectives.
Nice-to-haves- Experience developing models for and deploying models to run on edge-devices.
- Experience with deep learning training frameworks such as MLflow, Lightning, Weights and Biases, or similar.
- Experience with semantic mapping, image processing techniques and/or computer vision fundamentals.
- Experience with Apache Airflow (especially 3.x) and/or Astronomer.
- Experience with one or more of: Sensor Fusion, SLAM, Visual Odometry.
- 8+ years of professional experience
Benefits- Competitive salary and equity package
- Comprehensive medical, dental, and vision coverage for you and your eligible dependents
- Life insurance and short- and long-term disability coverage
- 16 weeks of fully paid parental leave to support all new parents
- Flexible, unlimited paid time off
- 401(k) retirement savings plan
- Free OneMedical membership
- Commuter benefits
- Snacks, goodies, and team lunches provided twice a week to keep you fueled and connected with your colleagues.
The estimated salary range for this position is $225,000 - $265,000 USD. Total compensation for this position is determined by skills, qualifications, relevant work experience, location, and other factors. This salary estimate excludes the value of any potential bonuses; the value of any benefits offered; and the potential future value of any long-term incentives. This information is provided per the New York City Human Rights Law. Please note that the range provided is applicable only to New York City-based applicants. Base compensation may vary if the work location is outside of New York City.
