Behavioural Science Consultant

About Samsung Food (Whisk)

At Samsung Food (you might know us as Whisk), we are pioneering the future of everyday care and health — helping individuals and families around the world live better through connected, proactive, and personalized experiences. Our mission is to connect food, health, and home across the Samsung ecosystem, turning data from your devices into personalized guidance that supports you and your loved ones.

We work at the intersection of AI, nutrition, behavior change, and digital health. The Samsung Food app was included in Google Play’s Best of 2020 Everyday Essentials list, has been regularly featured on the Apple App Store, and was nominated for a 2021 Webby Award. We’ve also launched Vision AI, which recognises food and meals using our global Food Genome, and developed deep integrations with Samsung Health and SmartThings.

We are a remote-first team of more than 100 people in over 30 countries, united by our commitment to innovation and impact. As part of Samsung Electronics, we combine the agility of a startup with the scale of the world’s largest consumer electronics company. Join us and help build technology that improves daily life for tens of millions of people across phones, wearables, TVs, and kitchen appliances in the years to come.

Learn more about how we’ve shaped a high-performing global team over the past 14 years.

Summary

Project Overview

We're looking for a behavioural-science specialist to take ownership of a set of currently unratified behavioural parameters inside an AI-powered weight management coaching engine, and to help shape the engagement mechanics (nudges, loop design) that sit on top of it. This is a fixed-scope consulting engagement structured around five defined deliverable milestones, not an ongoing support role.

You'll be working against a live backlog of flagged parameters across the engine's logic and content files, reviewing agent behaviour against COM-B and related behaviour-change frameworks, and helping design the daily/weekly engagement loops for a beta product surface. You'll work closely with our product managers and content team, and review material coming in from specialist freelance writers across a range of target behaviours. We have clinical reviewers on the team who handle medical safety sign-off directly — this is a behavioural science role, not a medical one: you're assessing whether the logic holds up against behaviour-change theory, not making clinical calls yourself.

Primary project deliverable: Close the provisional-parameter backlog to "no unowned parameters" in the shipped coaching engine, delivered as:

- A completed ratification register (item, decision, evidence, standing).

- Reviewed and edited the instructions that guide the engine's decision-making steps — working out a user's overall coaching approach, spotting the gap between where they are and their goal, deciding what to act on, and turning that into the message the user actually sees;

- A behavioural-guardrail evidence pass, confirming the product's safety-adjacent behavioural patterns are backed by real evidence;

- Flagging and framing anything that needs medical sign-off — the specific question, the options, and your recommended read — ready for the internal team to batch and route;

- An agreed pass bar and scored golden set for a set of currently uncalibrated evaluation criteria;

- Nudge/engagement-loop design recommendations for the beta surfaces.

- Behavioural sign-off on intervention contracts as they arrive from the freelance programme — barrier framing, and alignment with our internal behaviour-change model — across our full set of target behaviours.

Success metric/acceptance standard:

All seven deliverables above are complete: flagged items in the register carry a cited decision, a replacement value, or are flagged and framed for medical sign-off; the engine's decision-making steps have delivered, reviewed diffs; the evaluation criteria carry an agreed pass bar and a scored golden set; items needing medical sign-off are flagged and framed on a standing basis, ready for the internal team to batch and route; review turnaround on shipped engine changes is ≤2 working days, measured weekly; and all target-behaviour intervention contracts have received a behavioural sign-off (approved, sent back with named revisions, or escalated) within 5 working days of receipt.

Project Milestones

1. Finish reviewing the highest-priority behavioural rules and settings. Clear out any pending contract reviews. Flag and frame the first batch of items needing medical sign-off, ready for routing, by 15 Nov 2026.

2. Finish checking the behavioural guardrails. Agree on a scoring system for the review criteria and test it on sample cases, by 15 Dec 2026.

3. Finish reviewing the engine's decision-making steps. Clear all flagged items in the main settings files. Finish the engagement/nudge design work. Sign off on the remaining behaviour-contract reviews, by 15 Jan 2027.

4. Finish reviewing how the system performs with real (beta) users. Make sure nothing is left unreviewed, by 15 Feb 2027.

5. Update the review scoring system using real user data. Clear every review queue completely, by 15 Apr 2027.

What This Role Does Not Cover

- Final medical sign-off, and the batching/routing of flagged items to the clinical reviewer (owned internally) — you flag and frame what needs review, we handle the rest

- Content authoring (freelance specialists own first-draft writing)

- Final product or engineering decisions

Engagement Details

Structure: Hourly, with a weekly hours cap (max 20 hours). The five milestones are review points where we confirm progress

Duration: 6-month term

Location: Meaningful overlap with UK/CET working hours is preferred

What We're Looking For

- Strong grounding in behaviour-change theory (COM-B or equivalent), ideally with a research background

- Demonstrated experience evaluating or designing behavioural loops, nudges, or habit-forming product mechanics — not theory alone

- Comfortable working with structured, protocol-style content (reasoning rules, decision trees) rather than writing consumer-facing copy

- Experience working with cross-functional product/engineering teams on shipped features

Application Notes

Please share examples of behavioural frameworks you've applied to a shipped product feature.