Summer 2026
Research Engineer Intern
Hiverge
1-10 employees
AI platform for automatic algorithm design
No salary listed
Cambridge, UK
In Person
 AI & Machine Learning (1) |
|---|
People at Hiverge
People at Hiverge who can refer or advise you
- Masters degree in Computer Science, Engineering, related field, or equivalent experience.
- Strong coding skills in Python and experience with PyTorch/JAX and modern LLM tooling for inference and fine‑tuning.
- Experience in applied research engineering, demonstrated by impactful academic projects or internships in LLMs, search/RL, or compilers/verification.
- Exposure to evaluation frameworks, unit/property‑based testing, and performance profiling.
- Support the development of scalable experimentation foundations with strong reproducibility and observability
- Help create and improve robust prompt and code-selection tooling.
- Assist in optimizing our system’s inference and infrastructure for better efficiency.
- Work with the team (ML engineers and researchers) to implement new ideas.
- Engage with the research community to advance the state of algorithmic discovery.
- Contributions to open research or community tooling.
- Familiarity with algorithm design, and practical understanding of reinforcement learning, evolutionary methods and optimization methods.
Hive is a platform that automatically designs, tests, and optimizes algorithms for important business and machine learning tasks. It runs multiple agents in parallel, evaluates options against real performance metrics, and uses synthesis-style methods to assemble high-performance, task-tailored algorithms that adapt as environments change. Unlike generic methods, Hive delivers bespoke solutions without requiring large in-house engineering teams and stays effective as conditions shift. Its goal is to let organizations access custom, high-performance algorithms quickly and with less engineering effort, improving efficiency in planning, training, routing, and other critical tasks.
Company Size
1-10
Company Stage
Seed
Total Funding
$5M
Headquarters
Cambridge, United Kingdom
Founded
2025
People at Hiverge
People at Hiverge who can refer or advise you
Simplify's Take
What believers are saying
- Quantinuum uses Hive to translate natural language into quantum circuits, democratizing quantum algorithm design for researchers.
- Hive identified training algorithms surpassing human methods by over 15% on established AI benchmarks for resource allocation.
- Hive uncovered long-term planning algorithms operating magnitudes faster than traditional systems in an industrial challenge from Airbus.
What critics are saying
- Google DeepMind's AlphaEvolve delivers 23–32% speedups in LLM tasks, directly outcompeting Hiverge's core value proposition soon.
- Open-source AlphaTensor and FunShare replicate Hive's parallel-agent architecture at zero cost, eroding paid cloud marketplace adoption.
- Hiverge's six-employee team lacks engineering depth to scale beyond proof-of-concept while competitors deploy validated enterprise libraries.
What makes Hiverge unique
- Hiverge's Hive platform uses parallel agents to discover custom algorithms surpassing human baselines by over 15%.
- Former Google DeepMind researchers automate algorithm design, testing, and optimization for critical business and machine learning tasks.
- Hive adapts to concrete metrics, enabling businesses to accelerate AI training and optimize resource allocation without large engineering teams.
Help us improve and share your feedback! Did you find this helpful?