1.54M jobs liveBrowse

Data Scientist at Coefficient · Katori

Asks for

The ad says“Be excellent at agentic coding — comfortable driving tools like Claude Code, Cursor and GitHub Copilot, and skilled at reviewing and correcting their output (see "🤖 Agentic Coding is Core to How We Work" above).”
The ad says“Be comfortable with the building blocks of modern AI systems: embeddings and vector databases, prompt engineering, and working with frontier model APIs (OpenAI, Anthropic, Google).”
The ad says“Be excellent at agentic coding — comfortable driving tools like Claude Code, Cursor and GitHub Copilot, and skilled at reviewing and correcting their output (see "🤖 Agentic Coding is Core to How We Work" above).”
The ad says“Be excellent at agentic coding — comfortable driving tools like Claude Code, Cursor and GitHub Copilot, and skilled at reviewing and correcting their output (see "🤖 Agentic Coding is Core to How We Work" above).”
The ad says“Be comfortable with the building blocks of modern AI systems: embeddings and vector databases, prompt engineering, and working with frontier model APIs (OpenAI, Anthropic, Google).”
The ad says“Have hands-on experience with at least one LLM application pattern — Retrieval-Augmented Generation (RAG), fine-tuning, or building LLM evaluation suites — and care about measuring quality (hallucination/faithfulness, LLM-as-judge) rather than shipping on vibes.”
The ad says“Be excellent at agentic coding — comfortable driving tools like Claude Code, Cursor and GitHub Copilot, and skilled at reviewing and correcting their output (see "🤖 Agentic Coding is Core to How We Work" above).”
The ad says“Be comfortable with the building blocks of modern AI systems: embeddings and vector databases, prompt engineering, and working with frontier model APIs (OpenAI, Anthropic, Google).”
The ad says“Have hands-on experience with at least one LLM application pattern — Retrieval-Augmented Generation (RAG), fine-tuning, or building LLM evaluation suites — and care about measuring quality (hallucination/faithfulness, LLM-as-judge) rather than shipping on vibes.”
The ad says“Think about getting models into production — versioning, monitoring and iterating (MLOps / LLMOps), not just notebooks.”
The ad says“Be fluent in Python and SQL for data analysis, data science and machine learning, and comfortable across the Python Open Data Science Stack (pandas, NumPy, scikit-learn, matplotlib, Seaborn).”
The ad says“Think about getting models into production — versioning, monitoring and iterating (MLOps / LLMOps), not just notebooks.”
The ad says“Be fluent in Python and SQL for data analysis, data science and machine learning, and comfortable across the Python Open Data Science Stack (pandas, NumPy, scikit-learn, matplotlib, Seaborn).”
The ad says“Be comfortable with the building blocks of modern AI systems: embeddings and vector databases, prompt engineering, and working with frontier model APIs (OpenAI, Anthropic, Google).”
The ad says“Be fluent in Python and SQL for data analysis, data science and machine learning, and comfortable across the Python Open Data Science Stack (pandas, NumPy, scikit-learn, matplotlib, Seaborn).”
The ad says“Be comfortable with the building blocks of modern AI systems: embeddings and vector databases, prompt engineering, and working with frontier model APIs (OpenAI, Anthropic, Google).”
The ad says“Be fluent in Python and SQL for data analysis, data science and machine learning, and comfortable across the Python Open Data Science Stack (pandas, NumPy, scikit-learn, matplotlib, Seaborn).”
The ad says“Have hands-on experience with at least one LLM application pattern — Retrieval-Augmented Generation (RAG), fine-tuning, or building LLM evaluation suites — and care about measuring quality (hallucination/faithfulness, LLM-as-judge) rather than shipping on vibes.”
The ad says“Be fluent in Python and SQL for data analysis, data science and machine learning, and comfortable across the Python Open Data Science Stack (pandas, NumPy, scikit-learn, matplotlib, Seaborn).”
The ad says“Be fluent in Python and SQL for data analysis, data science and machine learning, and comfortable across the Python Open Data Science Stack (pandas, NumPy, scikit-learn, matplotlib, Seaborn).”
The ad says“Be fluent in Python and SQL for data analysis, data science and machine learning, and comfortable across the Python Open Data Science Stack (pandas, NumPy, scikit-learn, matplotlib, Seaborn).”
The ad says“Be comfortable with the building blocks of modern AI systems: embeddings and vector databases, prompt engineering, and working with frontier model APIs (OpenAI, Anthropic, Google).”

The Role in 30 Seconds

Full-time Data Scientist
Build and deploy cutting-edge AI and ML solutions for diverse clients (from Government to Startups).
Gain full-stack delivery experience across an array of industries while benefiting from investment in your professional growth and expertise.

Working at Coefficient

You'll be involved in a wide variety of projects, from cutting-edge AI solutions for the UK government to building transformative tools across a range of industries. This isn't just a standard Data Science position; you will gain hands-on experience by delivering end-to-end data science and engineering solutions for our clients, alongside building and improving our own internal products.

You can also expect plenty of mentoring and guidance along the way: we aim to be best-in-class at what we do, and we want to work with people who share that same attitude.

As a unique and fast-growing consultancy, this is an excellent opportunity to make a significant impact and shape our future success.

🚀 About Coefficient

Coefficient is a full-stack data consultancy dedicated to helping organisations solve their toughest challenges using data science, software engineering, machine learning, analytics, and artificial intelligence.

🔧Consulting & Delivery: We partner with clients to deliver end-to-end solutions, combining statistical expertise with agile delivery. This might involve developing cutting-edge models for a UK government agency, or working as an in-house team with a fast-growing tech start-up.
🎓 Training: Beyond consulting, we create and deliver tailored training programmes via workshops, online learning, and hybrid curricula to help our clients build their own internal skills. Past clients include BNP Paribas, EY, Hawk-Eye, the BBC, ACCA, CIOT, and the Metropolitan Police.

We enjoy variety in our work. One week, you might be developing high-speed trading algorithms; the next, you could be optimising logistics for delivery drivers or building election forecasting models.

👥 Our Team and Culture

Our team is our greatest asset. We invest heavily in professional development through our "10% Time" programme and our annual conference budget. We work with highly intelligent and passionate people who take pride in their work and enjoy a high level of independence.

🤖 Agentic Coding is Core to How We Work

We're unapologetically pro AI-assisted development. We expect our Data Scientists to be genuinely good at agentic coding — using tools like Claude Code, Cursor, GitHub Copilot, Windsurf and Codex to deliver faster without ever compromising on quality.

We treat this as a genuine, emerging discipline. Knowing how to specify work clearly, drive an agent through a task, and critically review and correct "almost-right" output is a real skill in itself — and one we actively invest in mastering. If you're already running spec-driven or multi-agent workflows, you'll feel right at home; if you're keen to get great at it, we'll help you get there.

Requirements

Our ideal candidate would:

Be fluent in Python and SQL for data analysis, data science and machine learning, and comfortable across the Python Open Data Science Stack (pandas, NumPy, scikit-learn, matplotlib, Seaborn).
Have hands-on experience with at least one LLM application pattern — Retrieval-Augmented Generation (RAG), fine-tuning, or building LLM evaluation suites — and care about measuring quality (hallucination/faithfulness, LLM-as-judge) rather than shipping on vibes.
Be comfortable with the building blocks of modern AI systems: embeddings and vector databases, prompt engineering, and working with frontier model APIs (OpenAI, Anthropic, Google).
Think about getting models into production — versioning, monitoring and iterating (MLOps / LLMOps), not just notebooks.
Be excellent at agentic coding — comfortable driving tools like Claude Code, Cursor and GitHub Copilot, and skilled at reviewing and correcting their output (see "🤖 Agentic Coding is Core to How We Work" above).
Enjoy sharing knowledge, experience, and passion with others, and have great communication skills — you'll write and contribute towards presentation slide decks to showcase our work during sprint reviews and client project demos.

We recognise that diverse teams are the most successful teams, and we know some people are less likely to apply for the role unless they are 100% qualified. Please do not worry if you don’t meet every single requirement listed.

We strongly encourage you to apply if this role excites you and you believe you have the potential to grow here. If you are unsure, please reach out to us - we would genuinely love to hear from you. We are committed to fostering a diverse, inclusive, and empowering culture at Coefficient.

📍 Location and Eligibility Requirements

This is a UK-based, hybrid role.

While we operate remotely for most of the month, we value in-person collaboration and regularly gather the whole team. The successful candidate must be able to travel to London for on-site work approximately 2-4 days per month.

Eligibility: You must already have the right to work in the UK.
Visa Sponsorship: Please note, we are unable to offer sponsorship for a Skilled Worker visa for this position.
Students: We are unable to consider applications from candidates currently in full-time education (including PhD students).

The Basics

📍 Location: We are based in Central London, but we are remote-friendly. You may be required to work on-site at clients’ offices.

💰 Salary: £38,000 annual salary with a meaningful uplift following a performance review at the successful 3-month probation mark.

🏖 Holiday: 33 days of annual paid holiday, including bank holidays.

💷 Pension: We're set up with Smart Pension to make sure we're contributing to help you save for retirement.

📈 Performance Reviews: Regular check-ins to ensure you’re progressing in your career and maximising your potential.

🚀 Opportunity: To be part of a unique and exciting company that prizes excellence of work. You will work closely with the CEO and become part of a dedicated and forward-thinking team. We want you to push yourself to learn new skills and be recognised as one of the best in your field.

♻️ Commitment: We were one of the first 80 signatories of TechZero. We are committed to challenging the status quo and are always looking for ways to make a positive impact.Our ideal candidate would:

Be fluent in Python and SQL for data analysis, data science and machine learning, and comfortable across the Python Open Data Science Stack (pandas, NumPy, scikit-learn, matplotlib, Seaborn).
Have hands-on experience with at least one LLM application pattern — Retrieval-Augmented Generation (RAG), fine-tuning, or building LLM evaluation suites — and care about measuring quality (hallucination/faithfulness, LLM-as-judge) rather than shipping on vibes.
Be comfortable with the building blocks of modern AI systems: embeddings and vector databases, prompt engineering, and working with frontier model APIs (OpenAI, Anthropic, Google).
Think about getting models into production — versioning, monitoring and iterating (MLOps / LLMOps), not just notebooks.
Be excellent at agentic coding — comfortable driving tools like Claude Code, Cursor and GitHub Copilot, and skilled at reviewing and correcting their output (see "🤖 Agentic Coding is Core to How We Work" above).
Enjoy sharing knowledge, experience, and passion with others, and have great communication skills — you'll write and contribute towards presentation slide decks to showcase our work during sprint reviews and client project demos.

Benefits

💸 Co-working Spaces: Regular co-working days at different locations in London with the team plus full access to the Hubble co-working network at all times to use a space near where you live.

🎓 Learning and Professional Development: Potential to improve skills through paid courses and subscriptions. We encourage all our team to engage with professional communities, we actively sponsor PyData Meetups and Humble Data, and we provide additional support for anyone wishing to speak at meetups/conferences.

🎟️ Conference Budget: £1000 per employee in year 1, rising to £2000 by year 3. This can help cover tickets, accommodation, and travel to attend relevant conferences.

👂 Spill: All-in-one mental health support programme with on-demand access to a variety of support. We cover 8 hours of therapy with a remote therapist for each team member every year, worth up to £520.

🧠 Headspace: Paid membership to Headspace to encourage good daily mental practices.

📚 10% Time: 4 hours per week dedicated to improving skills or pursuing your own project.

💻 Laptop & Peripherals: Company-owned Apple laptop plus peripherals such as a monitor and keyboard, for making remote working both comfortable and safe.

💃 Team Culture: We have a fantastic small team who enjoy socials together - everything from guided walking tours to escape rooms to Bake Off experiences.

📋 What to expect from the hiring process:

We aim for a transparent, efficient, and enjoyable hiring process. Here is what you can expect:

Round 1: Application Screening

We review your application materials (CV, screening questions, and code samples) to assess the initial match.
Note: Your application must include answers to the screening questions and code samples to proceed beyond this stage.

Round 2: One-Way Video Interview (Non-Technical)

This is designed for us to get a better sense of your interests and personality outside of your technical skills.

Round 3: Practical Coding Exercise (1 hour)

You will be booked for a 1-hour slot to complete a coding test. This exercise is carefully designed to mirror the practical, real-world data tasks you can expect to do at Coefficient.

Round 4: Technical Interview (1 hour)

You will meet with a member of our Data Team for a deep dive into the technical skills required for the role. Expect a collaborative session, including pair programming, to see how you approach problems in a team environment.

Round 5: Final Conversation with the CEO

This is an opportunity to discuss your motivations, long-term career goals, and ensure a strong cultural alignment. We want to know that you'll be a great fit for our team, but we also want to help you achieve your goals.

⏱️ Our Commitment to You

Speed: We are committed to moving quickly with this role, and you can expect swift feedback after each completed round.
Feedback Policy: We are unfortunately unable to offer feedback before Round 2. Feedback for subsequent rounds will always be provided if requested.

Please ensure that emails from our hiring platform (Workable) are not being filtered into your spam/junk folder. We want to make sure you receive all correspondence promptly!

Due to a large volume of applications, we are unable to consider applicants without code samples and submitted screening questions.