Therify

Therify

Senior Software Engineer

Mental health platform focused on connecting individuals with culturally responsive care providers

Worked on product and platform features that improved how users discover relevant content and get matched with the right therapists. The work sat at the intersection of personalization, trust, and reliability in a people-first product.

---

Key Outcomes

• Built AI-powered content recommendation system to improve user engagement and relevance.
• Developed AI-powered profile matching system for better therapist-user alignment.
• Improved reliability across onboarding and matching flows.
• Strengthened platform foundations for scalable personalization.

---

Client

Therify focuses on making mental healthcare more accessible and culturally aligned. The platform helps users find therapists and resources that match their identity, preferences, and lived experiences.

As the platform grew, the challenge was not just connecting users to care — but doing it in a way that felt relevant, trustworthy, and personalized at scale.

I worked on building systems that made those connections smarter and more reliable.

---

Contributions

My work primarily focused on AI-driven personalization and matching systems, along with improving the reliability of core user journeys.

1. AI-Powered Content Recommendation

I developed a recommendation system that surfaced relevant content based on user context and preferences.

• Designed logic to personalize educational and support content dynamically.
• Improved how users discover resources aligned with their needs and identity.
• Helped create a more engaging and supportive onboarding and post-onboarding experience.

---

2. AI-Powered Profile Matching System

I built a matching system that improved how users are paired with therapists.

• Translated user preferences, identity signals, and contextual inputs into matching logic.
• Improved match quality by aligning system outputs more closely with real user expectations.
• Supported iterative improvements based on feedback and behavioral signals.

---

3. Reliability in Core Flows

Beyond AI features, I contributed to stabilizing critical journeys:

• Improved onboarding and matching workflows where trust and clarity are essential.
• Reduced inconsistencies between frontend flows and backend responses.
• Helped ensure that personalization features didn’t compromise system reliability.

---

Challenges

1. Personalization vs predictability

AI-driven systems introduce variability, but user trust depends on consistency.

• Recommendations needed to feel personalized without being confusing or unpredictable.
• Matching needed to be flexible while still producing reliable outcomes.
• The challenge was balancing intelligence with clarity.

---

2. Translating human context into system logic

Mental health matching isn’t purely technical — it involves:

• identity and cultural alignment
• communication preferences
• nuanced user intent

Turning these into system inputs required careful abstraction without oversimplifying the problem.

---

3. Integrating AI into existing product flows

Adding AI features isn’t just about models — it’s about integration.

• Ensuring recommendations fit naturally into existing UI/UX flows.
• Making AI outputs explainable and usable for end users.
• Avoiding disruption to existing onboarding and matching experiences.

---

4. Maintaining reliability in a people-first product

This platform operates in a sensitive domain.

• Users rely on the system during vulnerable moments.
• Inconsistent behavior can quickly erode trust.

The challenge was ensuring that new intelligent features enhanced the experience without introducing instability.

---

Accomplishments

This project strengthened my ability to work on applied AI systems within real products.

• Learned how to integrate AI features into production workflows, not just prototypes.
• Improved how I design systems that balance personalization with reliability.
• Gained experience translating human-centered problems into scalable technical solutions.
• Contributed to a product where engineering decisions directly impact user trust and well-being.

---

Tech Stack

Frontend

• React

Backend

• Node.js
• Service-layer logic for matching and recommendations

Infrastructure

• AWS

Processing / Workflows

• Celery for background tasks and async processing

AI / Personalization

• Rule-based + model-assisted recommendation logic
• Matching algorithms driven by user inputs and behavioral signals

---

What I’d Do Next

If I were to continue building on this, I’d focus on:

• deeper personalization models based on longitudinal user behavior
• explainability in AI-driven recommendations and matches
• feedback loops to continuously improve matching quality
• analytics around care journey outcomes and engagement

The next step would be evolving the system from reactive matching → proactive, adaptive care guidance.