Heron Data (S20)
We categorise and label bank transaction data

Machine Learning Engineer

$70k - $105k1.00% - 2.00%

London, UK
Full-time
Any (new grads ok)
About Heron Data

We are Heron Data, a Y Combinator-backed technology start-up based in London, UK.

  • Our mission is to enable better financial services for all by accelerating fintech innovation
  • We are currently focused on unlocking the power of bank transaction data to help companies generate insights or build new features, e.g., enabling companies to make faster and more accurate lending decisions based on bank data
  • We are an established, humble and hungry team from diverse backgrounds and are backed by prominent investors
  • We launched in fall 2020, and since then have consistently grown revenue 10% week over week, have customers who use us every day of the week, and process well over 1M transactions daily

About the role

Skills: Machine Learning

We are Heron Data, a Y Combinator-backed technology start-up based in London, UK. We are looking for a Machine Learning Engineer to join our team and work with us on some really tough and fun problems relating with bank transaction data. More in-depth description below.

Interested? Reach out to [email protected]!

About us

  • Our mission is to enable better financial services for all by accelerating fintech innovation
  • We are currently focused on unlocking the power of bank transaction data to help companies generate insights or build new features, e.g., enabling companies to make faster and more accurate lending decisions based on bank data
  • We are an established, humble and hungry team from diverse backgrounds and are backed by prominent investors
  • We launched in fall 2020, and since then have consistently grown revenue 10% week over week, have customers who use us every day of the week, and process well over 1M transactions daily

The role

  • As a Machine Learning Engineer, you will be responsible for understanding and solving complex customer problems relating to bank transaction data
  • We envision that the primary way you could solve customer problems is by developing and deploying models, ranging from simple heuristic approaches to more advanced ML
  • For example, we have multiple classification models deployed that help our customers categorise bank transactions; these models utilise NLP, text vectorisation and few-shot learning, and are deployed as FastAPI / Flask microservices as part of a serverless architecture on GCP
  • You will also lead / contribute to designing the infrastructure needed to build, serve, and improve models at scale; this covers topics like experimentation, measurement, monitoring, and other tools to drive impact for customers
  • If you enjoy solving tough problems by deploying code, being part of a high calibre and tight-knit team, and seeing real impact from your work, this role could be for you!

What we offer

  • Ownership and autonomy over solving complex problems that are key to the success of the company
  • Freedom to use any approach to solve problems; this includes investigating various types of models, reading research papers, experimenting with and implementing solutions to make a positive impact on our customers
  • An outcome-focused, low-bureaucracy, engineering-first work environment with an emphasis on shipping code and high velocity engineering
  • A supportive and inclusive culture; quarterly retreats, monthly gatherings, weekly meals, and impromptu BBQ / hot pot hangouts that are always vegan friendly
  • London-based office with flexibility to work remotely. We 100% sponsor work visas
  • Substantial equity compensation (1.0-2.0%) and market-comparable cash compensation (£50k-£75k)

Requirements

We are looking for somebody who is proficient in inference, has basic software engineering skills, but above all enjoys making an impact by building and shipping code and models.

For example:

  • You thrive at solving tough problems, often with ML
    • You enjoy breaking down challenging problems into simple & well-designed solutions
    • You have knowledge of different inference approaches to solve real world problems - from simple heuristics to more advanced ML techniques
    • You are metrics-driven and want to measure the real impact of your work
    • (optional) You have a deep understanding of NLP and text classification methods
  • You enjoy building products and making real-world impact
    • You take responsibility over end-to-end deployment of your solutions — from ideation to development to deployment to improvement. e.g., you don't just live in Jupyter notebooks and you do know the basics of what an API is
    • You like shipping code frequently, iteratively, and collaboratively; You value MVPs and customer impact over perfectly engineered systems
    • You care about reproducibility and good MLOps, e.g., model versioning, experimentation
    • You choose the best framework / technology for the task, even if you might have favourites. We work primarily in Python but are always experimenting with what framework enables the greatest performance and delivers the most value
    • (optional) You enjoy solving ML infrastructure scale challenges and have deployed models that can handle large volumes and velocity of data
  • You like to work in an fast-paced, high-growth, team-based environment
    • You are entrepreneurial. This means having a growth mindset, finding problems without being told about them, thinking independently about issues and clearly advocating your position, and taking full ownership of your work, even if things go wrong
    • You are a team player. You value diversity & inclusion in the workplace, are a helpful teammate, and understand best practices for building products as part of a team e.g., code review / version control

Reach out to [email protected] with any questions!

Technology

We have 3 core components to our tech stack: backend, inference, and frontend.

  • Our backend is a Python/Flask API with postgres, redis, elasticsearch hosted on GCP/Heroku, handling well over 1M transactions per day
  • Our inference is based in DVC for local development and hosted on GCP as Cloud Runs. We use a variety of NLP, text vectorisers, and classification techniques to improve categorisation and merchant extraction for our customers
  • Our frontend is written in Svelte and deployed also as a GCP Cloud Run, and is currently used by our internal team but we plan to make this available for external usage as well

Other jobs at Heron Data

Machine Learning Engineer
fulltime
London, UK
Machine Learning
$70k - $105k
1.00% - 2.00%
Any (new grads ok)