Where human behavior
meets intelligent systems.
I translate complex human behaviors into intuitive digital experiences — and build AI agents and automation systems that feel natural to use. My foundation is in HCI and Psychology.
Location
Tel Aviv, Israel
Available
Freelance · Full-time
Languages
Hebrew · English
Specialty
Mental Health UX
Selected Work
UX Research · Mental Health · Mobile
Led user research for a mental health app. Leveraged qualitative insights to simplify the onboarding journey, reducing cognitive load and significantly cutting down early drop-off rates.
UX Research Lead · Healthcare · Data Visualization
Dual-view interface for physiological monitoring in misophonia treatment — serving therapists and patients from one data model.
Product Design · Emotional Support · End-to-End
Real-time emotional support platform for misophonia sufferers. From stakeholder interviews to Figma prototype — validated with 8/8 task completion.
AI System Design · Automation · Recruiting
Three-node Make automation: Fillout form → Gemini AI scoring → Gmail report. Structured candidate evaluation delivered in under 60 seconds.
Expertise
User interviews, usability testing, journey mapping, insight synthesis. Research that drives real product decisions.
End-to-end design: wireframes to high-fidelity Figma prototypes. Rooted in psychology and accessibility-first thinking.
Building agents with Claude API — context engineering, tool design, UX principles for conversational AI.
Complex workflow automation with Make and n8n. Translating manual processes into intelligent, scalable systems.
M.A. HCI, B.A. Psychology. Cognition, emotion, and behavior — the human layer every system needs.
Specialized experience designing for sensitive populations. Mental health apps, clinical interfaces, accessibility-first.
About
"Understanding human behavior is the prerequisite for designing good technology — whether that's a mobile app or an AI agent."
I'm a UX Researcher and Product Designer with a strong foundation in Psychology and Human-Computer Interaction. I specialize in translating complex human behaviors into intuitive digital experiences.
Beyond UX, I build — AI agents, automation workflows, recruiting systems. I don't separate the design work from the technical work. The best products come from understanding both.
Before transitioning into UX and AI, I spent 4.5 years as an IT/Communications Officer in the IDF — managing complex technological systems under pressure, where clarity and reliability weren't optional. I also worked for two years with at-risk youth and a year supporting children on the autism spectrum. That time on the ground shaped the way I think about inclusive design, edge cases, and building for people whose needs rarely make it into the default use case.
Reichman University
M.A. Human-Computer Interaction
Interaction Design, Cognitive Psychology
Ariel University
B.A. Psychology & Sociology
Dean's Honors. Graduated with distinction.
Netcraft Academy
UX Design Certificate
Contact
Whether it's UX research, an AI system, or automating a complex workflow — I'd love to hear about it.
Based in Tel Aviv — available for remote work globally. Response time usually within 24 hours.
Open to UX research contracts, full-time product design roles, AI consulting, and automation projects.
Led user research for a mental health and habit-building app. Leveraged qualitative insights to simplify the onboarding journey, reducing cognitive load and significantly cutting down early drop-off rates.
Problem
Angel had strong retention for active users — but was losing over 60% during onboarding. My role: lead the full research cycle from recruiting through synthesis into actionable design decisions.
Original onboarding — friction points
Process
10 in-depth interviews conducted by the research team with users aged 20–38. Semi-structured sessions focused on mental models of habit formation, emotional vulnerability, and decision fatigue.
10 participants — churned users, active users, first-timers.
40-min semi-structured sessions conducted by the research team. Think-aloud, recorded with consent.
8 participants through current onboarding. 3 critical drop-off points identified.
Thematic analysis and affinity mapping were used to synthesize raw data. We focused on identifying psychological barriers, specifically how 'Emotional Vulnerability' and 'Decision Fatigue' directly contributed to the 60% drop-off rate.
Insights
Personal mental health questions before any trust was built caused abandonment. Timing matters as much as content.
14+ choices in the first 3 screens. Users felt overwhelmed — the opposite of what they came for.
Users shown concrete app value before setup completion had 3× higher completion rates.
Solution
Build trust first, show value before asking, progressively disclose personal questions over the first week.
Redesigned flow — progressive disclosure
Results
Onboarding completion rate
Drop-off at question screens
Completion when value shown first
Designed an adaptive physiological monitoring interface to support misophonia treatment. One data model — two completely different interfaces for therapist and patient.
Context
Misophonia sufferers and their therapists need the same physiological data — but in completely different forms. Clinical tools are inaccessible to patients. Patient-facing tools lack clinical precision. The solution: one system, two views.
Process
6 sessions with therapists. Mapped clinical workflows and data needs.
8 patients, 2 weeks. Trigger events, emotional context, self-reports.
14 health monitoring dashboards reviewed for data visualization patterns.
Custom criteria: comprehension, stress response, task success per user type.
Design
Therapist view: raw physiological data with clinical controls. Patient view: the same data in calm visual language — zero medical jargon.
Therapist dashboard
Patient view — calm language
Results
User groups from one data model
Patient data comprehension
Anxiety in patient view testing
Real-time emotional support platform for misophonia sufferers. End-to-end product design: stakeholder interviews, competitive analysis, user flows, high-fidelity Figma prototype.
Problem
All existing support for misophonia is asynchronous — therapy, forums, guides. When a trigger hits, nothing helps in the moment. Counter Live fills that gap: real-time tools, community connection, coping support on demand.
Insights
10–15 seconds before a trigger escalates. Active support needed in 3 taps from any state.
Every competitor led with content. Users said peer connection in the moment felt more grounding.
When emotionally activated, complex UI fails. One action per screen. Decision overhead close to zero.
Design
Emergency flow — 3 taps to active support
Results
Taps to active support
Task completion in testing
Reported feeling supported
End-to-end recruiting system with four AI agents in n8n + Claude API. UX design principles applied to workflow automation — seamless for recruiters and candidates alike.
Architecture
Service design first — the full candidate journey was mapped before writing a single prompt. Then four AI agents were built to handle distinct pipeline stages.
Scores CVs against criteria. Outputs structured evaluation with plain-language reasoning.
Coordinates availability, sends personalized invites, handles rescheduling via Google Calendar.
Status updates in a warm, human tone — customized per candidate from application data.
Tracks pipeline in Airtable, flags stale stages, surfaces daily recruiter digest with action items.
UX Design
Every AI score includes reasoning. Every recommendation has an override. The human is always the final decision-maker.
Recruiter dashboard
Principles
Every score includes a plain-language explanation. Users understand why — critical for hiring decisions.
The system is a tool, not an authority. Recruiter judgment is always the final layer.
Automated messages feel personal. Excellent response times — without candidates knowing an AI was involved.
Results
Time on CV screening
AI agents in parallel
Candidate response rate
Three-node Make automation that turns a Fillout form submission into a Gemini AI-scored evaluation — structured report delivered to the recruiter's inbox in under 60 seconds.
Architecture
A lean, fully automated flow. No manual steps between submission and evaluation — the recruiter receives a structured report the moment the candidate clicks submit.
Make automation — live pipeline
5-step Hebrew candidate form — collects name, phone, email, CV upload, and simulation answers.
Receives form payload, evaluates each simulation answer, scores 0–100, generates plain-language reasoning.
Sends structured evaluation report to recruiter — score, brief, and final recommendation included.
Form Design
A 5-step Fillout form designed for clarity and low friction. Hebrew RTL layout, clean step-by-step flow — candidates know exactly where they are and what's left.
Fillout form — step 1 of 5 (candidate view)
Output
Gemini structures its output as a recruiter-ready report: candidate name, numerical score, plain-language explanation, and a clear final recommendation. Zero ambiguity for the hiring manager.
Gmail report — recruiter receives this per submission
Design Decisions
Three nodes only. Every additional node is friction — for the builder and for the system. Lean by design.
Gemini prompt engineered to return consistent report format every time — score, brief, recommendation. No freeform noise.
5-step form breaks a complex submission into digestible steps. RTL Hebrew layout, clear field labels, no unnecessary friction.
Results
Node pipeline — fully automated
Evaluation delivered after submit
Consistent scoring across all candidates