P
P
Pointer is building the interface between humans and software. Our product lives inside other apps — integrating in minutes, crawling the UI, and reading docs to understand how the product works. Once embedded, it guides users, answers questions, and helps them get things done in real time. The vision is simple: software that understands what you’re trying to do, and just does it.
We’ve raised $5.6m+ from Amplify Partners and an amazing group of angels from Anthropic, OpenAI, and more. Now we’re looking for a small group of gritty, fast-learning engineers to help invent this new layer of the stack.
About Pointer
Pointer is building the interface between humans and software. Our product lives inside other apps — integrating in minutes, crawling the UI, and reading docs to understand how the product works. Once embedded, it guides users, answers questions, and helps them get things done in real time. The vision is simple: software that understands what you’re trying to do, and just does it.
We’ve recently raised our seed round from incredible investors (to be announced). Now we’re looking for a small group of gritty, fast-learning engineers to help invent this new layer of the stack.
About the role
You’ll own how Pointer thinks — the intelligence that powers our guidance, copilots, and predictions. You’ll work on everything from inference optimization to RL training loops, building systems that make Pointer smarter every week. This is a role for someone who thrives on turning messy real-world data into production-ready intelligence.
Why this is exciting
Pointer is one giant closed loop: observe → understand → guide → learn. You’ll be tightening that loop and making it smarter. You’ll be working on challenges like:
- Cutting latency to sub-500ms without sacrificing quality
- Designing RL and search-based systems that learn which guidance actually leads to successful user outcomes
- Building eval harnesses that run on both synthetic and live traffic, and actually predict real-world performance
- Building models that can reason over arbitrary UI structure
- Feeding cleaned behavioral data back into training loops to make Pointer more helpful and less noisy over time
What you’ll work on
- Design and run experiments to improve Pointer’s speed and accuracy
- Prototype and optimize models (LLMs, SLMs, embeddings) for latency and reliability
- Develop memory systems that improve Pointer context
- Instrument user behavior and feed the data back into training loops for continual improvement
You’ll be a good fit if
- You’re strong in Python and ML frameworks, particularly PyTorch
- You think in terms of evals and metrics and care about what they mean in production
- You’re excited by messy data and figuring out how to make it useful
- You like working across the stack: inference optimizations, data pipelines, fine-tuning, and model monitoring
- You have experience with RL, retrieval systems, quantization, distillation, and model infra
- You’ve published ML papers or contributed to OSS projects