AI & Emerging technologies

ML / Data Science Vibe Coder

The Decision Lab

Montreal · Quebec · Canada

Share this job: X Facebook WhatsApp Save job : ML / Data Science Vibe Coder
We're looking for an

ML/Data Science Vibe Coder

who will build LLM pipelines (multi-agent simulations, LLM-judge scoring, evals), automate data analysis for client projects, and ship internal data tools. Everything we do at TDL is guided by

SPICE:

S ocially conscious,

P ragmatic,

I nventive,

C atalytic, and

E vidence-based. You can read more about us and our core values

here . Why this role is important for TDL

We started The Decision Lab because we saw a big gap between what science & technology can offer and what is actually applied in the real world. We believe that decision science - spanning psychology, neuroscience, and AI - has the power to fill that gap. We are not just consultants; we are a team of tinkerers. As an

ML / Data Science Vibe Coder , you will sit at the intersection of machine learning, engineering, and behavioral science. This role is fully execution-focused: no papers, no long research cycles. You will build the LLM pipelines, evaluation systems and data tools that power our client work, and you will ship them fast. You will work alongside senior partners and behavioral scientists to take ideas from “napkin sketch” to “shipped” in record time, and help us push how much of the heavy lifting in decision science AI can take on. What you'll be working on

LLM Pipelines:

Build multi-agent simulations, such as synthetic participants responding to messages or interventions, so we can stress-test ideas before they reach real people. LLM-Judge Scoring & Evals:

Design LLM-as-judge scoring systems and eval harnesses that reliably tell us whether outputs are good, including calibrating judges against human ratings. Automated Analysis:

Automate data analysis for client projects - cleaning, modeling and reporting on survey, experimental and behavioral data so our consultants get answers in hours, not weeks. Internal Data Tools:

Ship tools our team uses every day, like pipelines that pull and query project data, lightweight dashboards, and workflows that connect our systems to LLMs.

Who you are

You are a trained ML or data scientist who has fully embraced AI-native building. You have the rigor to know when a result is real and the speed to ship it before the next meeting. The Rigor: You understand statistics and ML deeply. You know what an eval is actually measuring, you spot leakage, bias and noisy judges, and you won’t put a number in front of a client that you can’t defend. The Toolkit:

You build daily with

Claude Code ,

Cursor ,

Codex

or similar tools, letting them handle the boilerplate so you can focus on pipeline design and getting the results right. Anyone working manually won’t keep up. The Stack:

You write production

Python

(pandas, NumPy, scikit-learn or similar) , work comfortably with LLM APIs

(Anthropic, OpenAI) , know your way around

SQL , and use

Git

as a matter of course. The Mindset: You are comfortable with ambiguity. You can take a business question (“Can we predict how people will react to this message?”) and turn it into a working pipeline without needing a 20-page spec, making calls on the fly or quickly validating with a colleague. The Mission:

You care about the world. You want your skills to contribute to projects that improve health, education, and financial well-being. Requirements

Education:

A master’s degree (or higher) in data science, machine learning, statistics, computer science or a related quantitative field. Experience: 2+ years of industry experience shipping ML or LLM systems in Python. Research-only or academic work does not count toward this. Portfolio:

Links to GitHub repos, deployed tools, or write-ups of pipelines and evals you have built. We want to see how you structure real systems. AI Fluency:

Demonstrated daily use of AI coding tools and hands-on experience building with LLM APIs. Communication:

Strong proficiency in English. You must be able to explain technical results and limitations to non-technical behavioral scientists and clients. Autonomy:

A self-directed work ethic. Since this role is remote and output-based, you need to manage your own time and troubleshoot independently. About the role

Expected hours : 40h/week Remote:

Fully remote, based in the Philippines, working 8 AM to 4 PM ET (Montreal time). That’s 9 PM to 5 AM Manila time during EST, and 8 PM to 4 AM during EDT.

#J-18808-Ljbffr

Reference: WJ-3875_16351955 · Posted 17 Nov 2026, 01:25 AM

Find more on GigBlow

This role is listed on GigBlow for discovery and search. Hiring decisions and applications are handled by the employer or their chosen application system.

ML / jobs in other states

Browse jobs