Hippocratic AI

Small

Hippocratic AI is a healthcare technology startup developing large language models designed for safety-focused, non-diagnostic patient-facing applications. The company is known for building AI agents that handle tasks like patient outreach, chronic care management, and health coaching.

3 interview notes · updated Jul 2026

Hippocratic AI·Machine Learning Engineer·Onsite - Coding / Algorithms

Jun 2026

Hippocratic AI had me do a live coding round on an actual repo, not a toy sandbox, which was a different kind of pressure than I expected for an MLE role. The whole thing tested whether you could orient yourself fast, make a targeted change, write tests, and lean on AI tooling without letting it make a mess of things.

Given an unfamiliar live repository, how would you orient yourself quickly, locate the right area to change, implement a small feature or bug fix, verify it with tests, and explain the trade-offs you made, all within a strict time limit while using AI coding assistants appropriately?

“The part that tripped me up was the orientation phase.”

View Post

Hippocratic AI·Software Engineer·Take-home Assignment

May 2026

Take-home assignment for a software engineering role at Hippocratic AI. The task was building an AI agent staffing scheduler from a CSV input, outputting hour-by-hour coverage needs for a full Pacific-time day. Pretty open-ended with a CLI requirement and an optional UI stretch goal.

Given a CSV of patient call requirements, build a scheduler that computes hour-by-hour AI agent staffing needs, broken down both in total and per customer, for a single Pacific-time day. A CLI demo is required; a UI is a bonus.

“The open-endedness was the hardest part.”

View Post

Hippocratic AI·Machine Learning Engineer·Onsite - System Design / Architecture

May 2026

Whiteboard system design round at Hippocratic AI for an ML Engineer role. The whole session was basically one giant deep-dive into LLM inference infrastructure, and they pushed hard on the follow-up scenarios rather than letting you coast on the happy path.

Design a high-throughput multi-GPU LLM inference serving system. Walk through the full stack including KV-cache and VRAM management, memory pool layout, fragmentation handling, and how you reuse memory across requests.
How do you handle multi-GPU communication bottlenecks in tensor parallelism? Specifically, how do you overlap computation with All-Reduce, and what's the role of custom fused kernels here?
An extremely long prompt arrives and exhausts your pre-allocated KV-cache pool. How do you handle this without crashing the node? Discuss CPU offloading, request preemption, recomputation, and admission control.

“This is where I spent most of the time.”

View Post