Case Studies

Role: Founder, AI Engineer, Full Stack Builder

Scope: Conceived, architected, and currently building a next-generation platform that replaces static resumes with dynamic, conversational AI profiles. Integrated AI interviews help extract values, passions, and goals from candidates. Recruiters can query the resulting avatar before initiating real-world contact.

Tech Stack: Python, LangChain, LangGraph, Chroma, OpenAI, n8n (aux), MCP (planned)

Overview: Satisph.ai is a next-generation AI-powered platform designed to replace static resumes with dynamic, interactive, conversational profiles. It's built to solve one of the most frustrating mismatches in modern hiring: the broken signal-to-noise ratio that prevents great candidates from finding the right opportunities, and vice versa.

The Insight: Traditional resumes are outdated. They're flat, lifeless, and fail to capture the richness of a person's story, values, or potential. Satisph.ai flips the script. Instead of skimming through pages of generic bullet points, recruiters can now interview a candidate's AI-powered avatar. And instead of tweaking resumes for algorithms, candidates get to speak their truth through a deeply personalized, intelligent interview experience with an AI.

AI-First Product Design: Satisph.ai is designed as an AI-native experience from the ground up. Resume owners are interviewed in-depth by a custom AI agent that doesn't just capture work history—it explores personality, passion, and values. The final product is an interactive, chat-ready profile that allows recruiters to query and converse with the candidate's experience in natural language. No scheduling, no cold calls. When a recruiter is ready to go deeper, the system facilitates a live conversation.

End-to-End Builder Mindset: From concept to deployment, I own every aspect of the build: defining the product vision, architecting the system, and building the AI pipeline using LangChain, Chroma, and Python. I'm using tools like Cursor, ChatGPT, and n8n for orchestration and rapid prototyping, blending product intuition with full-stack execution.

Token-Aware Model Selection: Different tasks are routed to different LLMs based on performance and cost needs: GPT-3.5 handles high-volume Q&A, while GPT-4 is reserved for narrative synthesis. This approach minimizes costs without sacrificing quality and is built to scale.

Modular Architecture: LangChain orchestrates stateful AI logic, and Chroma serves as the vector store for embedding and retrieval. The architecture is flexible and ready for future migration to Pinecone, Weaviate, or Azure if needed.

MCP Integration (Planned): The system is designed to support the Model Context Protocol, enabling advanced agentic behavior and tool orchestration in future versions.

User Flow Summary:

1. Resume owner uploads a resume
2. AI creates a personalized interview plan
3. AI conducts a deep interview
4. Responses are distilled into structured, narrative output
5. Owner approves and publishes an interactive profile link
6. Recruiters sign up, go through onboarding, and chat with the profile
7. If there's interest, the system facilitates a real conversation

Success Metrics: Delivering value to resume owners through intelligent interviewing, enabling recruiters to screen candidates meaningfully and efficiently, and facilitating direct engagement only when it truly matters.

↑ Back to Index

Role: Creator, Architect, Full Stack Engineer

Scope: Designed and built a fully autonomous AI media pipeline that transforms daily news headlines into lyrical songs, renders them as videos, and uploads them to YouTube – without human input.

Tech Stack: n8n, OpenAI GPT-4, FFmpeg, Laravel, Horizon, Suno/Udio (planned)

Overview: BoomBoom News is a fully automated, AI-driven media pipeline that transforms daily news headlines into original musical video content. By combining LLMs, generative music, video rendering, and automation tools, the platform turns real-time information into compelling creative output with zero human input after setup.

The Spark: With the rise of generative AI, the question emerged: what if the news didn't just inform us — what if it moved us? BoomBoom News explores the intersection of information, automation, and creativity. It takes an RSS feed of headlines, distills their emotional or narrative content using a large language model, and converts them into musical storytelling videos ready for YouTube.

How It Works:

1. Input & Interpretation: RSS feeds are ingested daily and GPT-4 generates lyrics inspired by the headlines.
2. Audio Generation: Lyrics are passed to a generative music model (originally Suno, Udio pending) to produce a track.
3. Video Rendering: FFmpeg combines the song with visuals, overlays, and effects. GPT-4 also creates YouTube metadata.
4. Automation & Publishing: n8n orchestrates the process; Laravel + Horizon queues and publishes the final product to YouTube.

Stack & Tools:

• n8n: Automation, retries, scheduling
• FFmpeg: Custom video rendering
• OpenAI GPT-4: Lyric and metadata generation
• Suno/Udio (planned): AI music generation
• Laravel + Horizon: Job queues and publishing automation

Results:

• Fully autonomous creative publishing system
• Designed to scale horizontally (different topics) and vertically (deeper story focus)
• Currently runs a daily news summary using five randomly selected stories
• Architecture is reusable for future generative media experiments

Why It Matters: BoomBoom News isn't just a cool demo — it's a blueprint for automating narrative creativity. Whether used for journalism, activism, or entertainment, it shows how raw data can be transformed into emotionally resonant content using a thoughtful AI stack.

Key Themes:

• End-to-end AI media pipeline
• Multi-modal AI usage: text → music → video
• Automation-first architecture
• Modular, fault-tolerant design
• Real-time transformation of information into storytelling

↑ Back to Index

Role: Founder, Product Designer, Full Stack Engineer, Customer Success

Scope: Designed and evolved a unique writing platform that helps users bypass their inner critic and enter deep flow states. Rebuilt the application from the ground up with a modern stack, while preserving over a decade of user writing data.

Tech Stack: Laravel, Vue.js 3, MySQL, Redis, Inertia, Alpine.js, Tailwind, OpenAI, Horizon, Spark, Nova

Overview: ILYS (I Love Your Stories) is a distraction-free writing platform designed to bypass the inner critic and help users get into creative flow. It allows people to write one letter at a time, without the ability to edit, delete, or even see more than the last character typed. This radical constraint helps users drop into a deep, judgment-free state of expression – ideal for journaling, storytelling, healing, and creative breakthroughs.

The Origin: Born from personal exploration and a desire to unblock creativity, ILYS began as a tool to help people express themselves more authentically. Over time, it evolved into a full-fledged SaaS with a growing international user base that includes writers, therapists, educators, and everyday people seeking to let go of perfectionism and reconnect with their voice.

How It Works:

1. Set a word count or time-based goal
2. Write – one character at a time, without the ability to act on self-doubt
3. Complete the session to unlock your creativity

By removing the ability to revise midstream, ILYS suspends self-judgment and fosters uninterrupted flow. Many users report feeling relief, catharsis, insight, and joy after their sessions.

Platform Features:

• Session-based writing architecture with nested Stories, Sections, and Sessions
• Full account system with onboarding and membership
• AI typo cleanup and reflective chat
• Privacy-first, ad-free design
• Monthly challenges, writing tutorials, and a playful UI

Behavior-Shaping Design: ILYS uses carefully tuned positive and negative reinforcement to help users practice "letting go" in a safe, consequence-free, and deeply supportive environment.

• Negative reinforcement: No delete or backspace. Attempts trigger subtle feedback (audible buzzers, screen flashes) to build awareness of habitual control.
• Positive reinforcement: Celebrations, progress bars, and a dopamine-rewarding writing flow encourage trust, release, and emotional safety.

This creates a sandbox where users can scream, confess, imagine, or dream — without fear of being judged, even by themselves. It's not just a text box; it's a practice ground for surrender, healing, and creative freedom.

Results:

• Millions of words written by users since 2015
• High long-term retention with 8+ year active members
• Widely reported boosts in mental clarity, creative confidence, and emotional release
• Used by authors, therapists, coaches, educators, and dreamers worldwide

Why It Matters: In an era of over-editing, comparison culture, and distraction, ILYS makes it safe to be unpolished. To be real. It's not about producing — it's about processing. Users leave sessions not just with words, but with insight, calm, and a deeper connection to themselves. The emotional safety of ILYS is its true power — and what makes it unlike any other writing tool on Earth.

Key Themes:

• Flow-state-first product design
• Behavior-shaping constraints to unlock authenticity
• Writing as healing and transformation
• High-empathy UX for creative courage
• Long-term community-driven growth

↑ Back to Index

Role: Full Stack Engineer via Atkins Professional Services

Scope: Contributed to the design, development, and scaling of key ordering and delivery systems for global restaurant brands.

Tech Stack: .NET, C#, SQL Server, JavaScript, jQuery, Cordova, AWS, Jenkins, PhantomJS

Overview: Tillster is a global leader in digital ordering and customer engagement solutions for the restaurant industry. Their platform supports web and mobile ordering, kiosk experiences, loyalty programs, and delivery integration. I worked with them for over five years through Atkins Professional Services, contributing across the SDLC on high-impact systems powering millions of transactions worldwide.

Core Systems:

• Order Manager: The central system for Tillster’s ordering infrastructure. I helped create workflows, menu logic, and reporting features from the ground up, scaling the system for use across global franchises.
• Delivery Manager: Integrated with third-party services (like Uber Eats and DoorDash) to manage real-time delivery coordination. I also built a Cordova-based mobile proof-of-concept to explore mobile-first deployment.
• ORB Kiosk Manager: Supported in-store multilingual kiosk interfaces with hardware integrations and custom flows per brand.

Experimental Builds:

• Tillster Testing Framework: I envisioned and created a modular, headless UI automation and reporting suite. This allowed QA to run multi-brand tests with real-time reporting and full CI/CD integration, improving stability across releases.
• Contributed to R&D and internal tooling, including architectural proof-of-concepts and performance tuning experiments.

Challenges:

• Scaling complex, multi-tenant systems across hundreds of brands with unique workflows and configurations
• Creating robust QA tooling that improved reliability without slowing development velocity
• Maintaining performance and security under tight enterprise delivery deadlines

Why It Matters: Tillster powers digital infrastructure for some of the world’s largest restaurant brands. The systems I helped build handled millions of daily transactions and shaped how customers order, pay, and engage across platforms. My contributions helped ensure these systems were robust, scalable, and adaptable to the evolving needs of modern dining experiences.

Key Themes:

• Multi-tenant SaaS architecture
• Enterprise-grade QA automation
• Full-cycle SDLC participation
• Performance and reliability at global scale
• Customer-first design in high-volume systems

↑ Back to Index

Role: Consultant via Cygnusoft

Scope: Integrated .NET systems into SAP workflows, built handheld warehouse apps, and created multilingual supply chain tools.

Tech Stack: .NET, SQL Server, SAP B1 Integration

↑ Back to Index

Role: Developer

Scope: Built features, designed schemas, migrated data, and mentored junior devs in a globally distributed team.

Tech Stack: ASP.NET, JavaScript, SQL Server

↑ Back to Index

Role: Lead Developer

Scope: Designed and delivered a platform for registration and management of all after-school programs in the district.

Tech Stack: PHP, MySQL, JavaScript

↑ Back to Index

Role: Consultant via Cygnusoft

Scope: Built customized messaging and reporting systems tailored to legal workflows.

Tech Stack: .NET, SQL Server

↑ Back to Index