About us

We exist to simplify healthcare and improve lives by making care feel instant, intelligent and human.

HealthHero is Europe’s largest digital health provider, delivering 4 million consultations per year. But we’re just getting started. We’ve built a seamless digital clinic that brings body and mind together — from GP appointments and mental health support to long-term condition management. By sitting behind the world’s leading insurers and employers and supporting public health systems, we make it easier for millions of people to get the care they need, exactly when they need it.

We are a high-growth, capital-backed business with a sophisticated scale strategy. Our team is a unique blend of digital-native pioneers, management consultants, creatives and industry-leading clinical experts.

We aren’t just digitising appointments; we’re building the next generation of healthcare. We’re creating an AI-powered, always-on ecosystem that learns from every interaction to shift the needle from reactive treatment to proactive, sustainable health.

At HealthHero, we are digital when it should be and human where it counts. Join us, and help build a next generation health system the world is waiting for.

Context Management & Decisioning for Agentic Systems

Minimum of 6 months I 2-3 days per week

Context

HealthHero is Europe's largest digital-first healthcare provider, with a library of clinically governed symptom-assessment algorithms developed over two decades with focused clinical and software development teams. We are now developing our agentic layer across a variety of use cases.

Maintaining coherent, trustworthy conversational state across a multi-turn clinical dialogue — and knowing when the system does not know something — is the subject of this position.

Research directions:

The successful candidate will shape their own project within this space. Indicative open problems:

• Long-horizon conversational state: Maintaining coherent structured state across extended dialogues.

• Managing/Calibrating uncertainty without model internals in agentic systems: Estimating extraction confidence when self-reported LLM confidence is uncalibrated and token-level probabilities are unavailable: consistency sampling, feature-based calibration, and lightweight domain embedding models for answer matching.

• Bounded autonomy in safety-critical agents: architectures that provably confine a language model to understanding rather than deciding, and verification of that boundary under adversarial input.

• Evaluation of multi-turn stateful agents: regression and robustness methodologies for systems whose correctness is a property of an entire conversation rather than any single response.

What the position offers:

• Real multi-turn clinical dialogue as the empirical setting, grounded in a clinically governed algorithm corpus and thousands of structured question–answer pairs.

• A direct route from research results into a regulated, patient-facing product — including shaping how emerging regulation (EU AI Act) is met in practice for conversational AI.

• Co-supervision between [academic supervisor] and HealthHero's technical leadership; support for publication and conference attendance.

Candidate profile:

Essential:

• Strong background in computer science, machine learning, computational linguistics or a related discipline (Master's level for PhD entry; PhD in a relevant area for postdoctoral appointment).

• Working knowledge of modern NLP and large language models. Structured information extraction, prompting and evaluation, and their characteristic failure modes.

• Grounding in probabilistic reasoning or uncertainty quantification (calibration, sampling-based methods, or related).

• Programming proficiency sufficient for empirical work on real systems (e.g. Python).

Desirable:

• Experience with dialogue systems, conversational state tracking, or agent architectures.

• Interest in AI safety, robustness or adversarial evaluation; familiarity with clinical NLP (e.g. SNOMED-CT/UMLS) is welcome but not required.

• Experience collaborating with industry or working with production datasets.

Closing date: Monday 21 September (9am)

Pay rate: competitive market rate

Equality, Inclusivity and Diversity

In line with our commitment to Equality, Inclusivity and Diversity, we welcome and encourage applications from all suitably qualified candidates from all backgrounds. We are committed to supporting and promoting equality and diversity and aim to establish an inclusive working environment. As such, we welcome diverse applications from candidates irrespective of age, disability, gender reassignment, marriage and civil partnership, pregnancy and maternity, race (including colour, nationality, ethnic and national origin), religion or belief, sex, or sexual orientation.

We are a certified Disability Confident Employer and are committed to affording equal opportunities for candidates with disabilities or special needs. Should you require any reasonable adjustments to be made at any part of your application process, please let us know by contacting us at recruitment-team@healthhero.com