Software Engineer, Multimodal Backend Systems

Eventual•San Francisco, CA

2d•Onsite

About The Position

Eventual is building the infrastructure to find any situation described across a fleet's entire video history, and turns it into a training set or an alert someone can still act on. Our open-source engine, Daft, is purpose-built for multimodal AI. We fine-tune and run the vision models ourselves, which makes indexing every hour cheaper than annotating a sample. We've raised $30M from investors like Felicis, CRV, Y Combinator, and angels from the co-founders of Databricks and Perplexity. Our team comes from AWS, Lyft, and Tesla. We powered the last generation of Physical AI in self-driving; now we're doing it for the next. Join our small (but powerful!) team, 4 days/week in our SF Mission District office. Our Mission: Our goal is to build Scenario Mining and Data Curation for robot fleet data. We empower Physical AI and robotics teams to instantly find, curate, and stream the data they need to train frontier models. Eventual is an agile team where every engineer has high ownership across the stack from our compute infrastructure, to our data storage/querying layers and model training/deployment.

Requirements

Strong engineers who are problem-solvers at heart
Excellent coding and architectural fundamentals in languages like Rust, C++, Python, or Go
A drive to reach for lower-level primitives when performance and efficiency demand it.

Responsibilities

Build our Streaming Architecture: Design, build, and optimize real-time WebRTC media pipelines and custom signaling mechanisms to stream multi-camera video feeds from robots to our platform
Manage Hardware-Accelerated Video Pipelines: Integrate and tune video codecs (H.264, HEVC/H.265, AV1) leveraging GPU acceleration (NVENC/NVDEC) for hardware decoding and dynamic bitrate adaptation
Scale Video Ingest & Storage: Engineer high-throughput video ingestion and distributed transcoding services that compress, index, and write petabyte-scale media corpora to object storage (AWS S3, GCS) for long-term retention and retrieval.
Build Spatial Data Pipelines: Construct specialized storage formats, spatial indexes, and processing workflows to ingest, align, and query dense 3D LiDAR point clouds, depth maps, and multi-camera spatial datasets.
Stream High-Frequency Telemetry at Scale: Develop high throughput ingestion pipelines capable of capturing, parsing, and storing real-time sensor streams and high-frequency robot state telemetry across our customers’ fleets.
Ensure Sensor-to-Video Synchronization: Coordinate time-sync protocols (PTP/NTP) across disparate sensor feeds to temporally align LiDAR point clouds, IMU telemetry, and video frames into unified data structures for downstream consumption.
Develop Data-Dense User Interfaces: Architect intuitive, high-performance web applications using React, TypeScript, and modern state management to handle continuous streams of data.
Visualize 3D Spatial & Video Streams: Build custom frontends to render 3D LiDAR point clouds, spatial bounding boxes, and multi-camera video feeds.
Build Agentic/AI-Native Product Workflows: Create agentic data-exploration, curation and search tools that empower robotics operators and AI researchers to review, annotate, and analyze complex physical AI datasets.