Working at Four Vectors
We are a small team in Noida building robots that understand the people around them. Every role below works on the same problem from a different side: (1) recording what people do, see and believe, (2) turning those recordings into data a robot can learn from, and (3) building the models that use it.
| Position | Focus | Terms |
|---|---|---|
| Computer vision engineer | Perception for our recordings | Full-time, Noida |
| Robotics full-stack engineer | From the recording rig to a trained policy | Full-time, Noida |
| Research scientist, mental models | Inferring what people believe | Full-time, Noida |
| Outreach manager | Participants, places and collaborators | Full-time, Noida |
| Study coordinator | Running recording sessions | Full-time, Noida |
1 Computer vision engineer
We record people at work with fixed cameras, eye-tracking glasses and a lightweight wearable. We need an engineer who turns those recordings into data a robot can trust: where each person is, where they look, and where every object is.
The work.
- Calibrate fixed multi-camera rigs, eye-tracking glasses and wearable cameras, and keep them calibrated.
- Put every stream on one clock: cameras, gaze, motion and audio.
- Map gaze from the glasses into the shared scene, so each look lands on an object.
- Track people, hands and objects through rooms the cameras only partly see.
- Build the quality checks every file passes: blur, occlusion, dropped frames and drift, and the face blurring every release goes through.
- Pre-label what happens, so annotators confirm instead of starting from zero.
- Export data that is ready to train on: MCAP, LeRobot v3, RLDS or HDF5.
You bring.
- Two or more years building computer vision for real-world video, not only benchmarks.
- Strong Python, PyTorch and OpenCV.
- Camera models, calibration and multi-view geometry you can explain on a whiteboard.
- A pose, detection, tracking or segmentation model you have taken from paper to production.
- Comfort with large video datasets, ffmpeg, Linux and GPUs.
Good to have.
- Eye tracking, stereo depth, visual-inertial odometry or SLAM.
- Egocentric or hand-object datasets, such as Ego4D or EPIC-KITCHENS.
- Papers, open-source work or code we can read.
Terms. Full-time, at our office in Noida.
2 Robotics full-stack engineer
Our recordings start in a room full of sensors and are meant to end in a robot’s policy. We need the engineer who owns everything in between. Experience with reinforcement learning is preferred.
The work.
- Own the recording software: sensor drivers, capture, and time sync across cameras, eye trackers and wearables.
- Build the pipeline from capture to dataset: ingest, encrypted storage and processing jobs at scale.
- Export clean episodes in MCAP, LeRobot v3, RLDS and HDF5, and check that they load on the other side.
- Show that the data works: train and evaluate policies on it, from imitation learning to reinforcement learning, simulation first.
- Build the web tools the team runs on: session tracking, quality review and labelling.
- Work with collaborating labs on formats, schemas and what their models need.
You bring.
- Two or more years shipping software across the stack, with robots or machine learning in the mix.
- Strong Python, plus C++ or TypeScript.
- ROS 2, sensor data and time synchronisation.
- PyTorch: you have trained models and read the loss curve honestly.
- A backend and a frontend you built and kept running (APIs, Postgres, React or similar).
- Linux, Docker and the cloud.
Preferred.
- Hands-on reinforcement learning: PPO, SAC or similar, reward design, sim-to-real.
- Isaac Sim or Isaac Lab, MuJoCo, or similar simulators.
- LeRobot, robot arms or embedded hardware.
Terms. Full-time, at our office in Noida.
3 Research scientist, mental models
We infer what a person believes about the world from what a robot can observe. We need a scientist to take that further: better models of what people perceive, measured against what they actually believe.
The work.
- Develop methods that infer a person’s belief state from cameras, gaze and the robot’s own observations.
- Model what people perceive from gaze, rather than assuming they see everything in view.
- Design the studies and spoken questions that record what participants believe, and evaluate methods against them.
- Write up results as technical reports and papers.
You bring.
- A PhD, or a master’s degree with publications, in robotics, human-robot interaction, computer vision or machine learning.
- Experience with probabilistic state estimation, world models, or models of human perception and behaviour.
- Strong Python and PyTorch, and experiments others can reproduce.
- Clear technical writing.
Good to have.
- Research on situation awareness, theory of mind or human factors.
- Studies with human participants, including consent and study design.
Terms. Full-time, at our office in Noida.
4 Outreach manager
Every recording starts with a yes. Before a session runs, someone has to find the people and places to record, explain the work and earn their trust. Then keep a full calendar running on time.
The work.
- Find and recruit participants, homes and workplaces for our studies.
- Explain the work clearly, on the phone and in person, to participants, owners and managers.
- Get consent and paperwork right, every time.
- Plan sessions end to end with the study team: people, places, equipment, travel and schedules.
- Build relationships with research labs and programmes, take sharp notes on calls and follow up quickly.
- Build the simple systems (sheets, checklists, a CRM) that let a small team do twice as much.
You bring.
- Clear, confident spoken English and Hindi. You can explain robots to a homeowner, and homes to a robotics researcher.
- Two or more years in field operations, partnerships, business development or community outreach.
- An operator’s habits: lists, follow-ups, deadlines kept.
- Spreadsheets you are proud of.
- Happy to travel across Delhi NCR when a study needs it.
Good to have.
- Experience recruiting people for research or surveys.
- A regional language beyond Hindi.
Terms. Full-time, at our office in Noida.
5 Study coordinator
Our recording sessions are short, closed and carefully run. We need someone to run them: set up the room, brief participants, ask the questions on time, and make sure every file is complete and safe.
The work.
- Set up and check the fixed cameras, eye-tracking glasses and wearable before each session.
- Brief participants, take written consent, and keep sessions closed to bystanders.
- Ask the short spoken questions on schedule and log what happened.
- Check recordings after each session and move them to encrypted storage.
- Keep equipment charged, calibrated, labelled and ready.
You bring.
- Experience running studies, field sessions, lab operations or events.
- Calm with people and careful with equipment.
- Clear spoken English and Hindi.
- Comfort with cameras, laptops and moving large files.
Good to have.
- Research with human participants, including consent procedures.
- Basic Python or other scripting.
Terms. Full-time, at our office in Noida.
6 How to apply
Write to us with the position in the subject line. Send a CV and one piece of work you are proud of: a repository, a paper, a project or a spreadsheet. We welcome applications from everyone.