Float.ai
Float.ai
AI-powered overlay tool that puts a teleprompter, sticky notes, and Claude AI on top of whatever you're doing — video calls, recordings, presentations. The desktop app creates a true click-through transparent overlay that floats above every window. Collapses to a tiny floating pill when you don't need it. Global hotkeys work even when another app has focus. Browser version lets you share your screen as the background so your audience sees your desktop while you read your script. Web notes with full browser engine (no iframe restrictions), draggable/resizable color-coded notes with opacity control, and an inline Claude chat that helps you mid-presentation. Built with Electron for desktop, vanilla HTML/CSS/JS for browser, deployed on Vercel.
AI-powered overlay tool that puts a teleprompter, sticky notes, and Claude AI on top of whatever you're doing — video calls, recordings, presentations. The desktop app creates a true click-through transparent overlay that floats above every window. Collapses to a tiny floating pill when you don't need it. Global hotkeys work even when another app has focus. Browser version lets you share your screen as the background so your audience sees your desktop while you read your script. Web notes with full browser engine (no iframe restrictions), draggable/resizable color-coded notes with opacity control, and an inline Claude chat that helps you mid-presentation. Built with Electron for desktop, vanilla HTML/CSS/JS for browser, deployed on Vercel.
Product.ai
Product.ai
product.ai is an AI-powered product management workspace where six specialized agents, competitive intel, market research, feedback triage, PRD writer, sprint planner, and design brief handle the heavy lifting of product work. Agents chain together in automated pipelines: one click runs a full product analysis end to end, with each agent's output feeding context into the next. Built with React 19, Groq (Llama 3.3 70B), and Tavily for real-time web research, the app includes a markdown document editor with version history, a drag and drop kanban roadmap, inline commenting, a decision log with AI extraction, and integrations with Linear and Slack.
product.ai is an AI-powered product management workspace where six specialized agents, competitive intel, market research, feedback triage, PRD writer, sprint planner, and design brief handle the heavy lifting of product work. Agents chain together in automated pipelines: one click runs a full product analysis end to end, with each agent's output feeding context into the next. Built with React 19, Groq (Llama 3.3 70B), and Tavily for real-time web research, the app includes a markdown document editor with version history, a drag and drop kanban roadmap, inline commenting, a decision log with AI extraction, and integrations with Linear and Slack.
Prototype for Canva
Prototype for Canva
Designed and shipped the Consistency Agent, an AI system that catches when your marketing campaign says different things across different formats. Brand Kit makes sure your designs look the same. The Consistency Agent makes sure they say the same thing. It scans every asset in a campaign using a three-tier taxonomy: Must-Match (CTAs, dates, prices, URLs), Should-Match (tone, hero image), and Format-Dependent (layout, text length). One-click Sync All propagates the correct version everywhere. Built a full working prototype, campaign dashboard with 8 asset types, real Claude API integration with streaming, live drift detection in the canvas editor, brief upload with canonical value extraction, and a 9-step guided onboarding tour. AI architecture uses semantic role classification, sentence embeddings (MiniLM), perceptual hashing, and rule-based propagation. Spec included 24-week ship plan, A/B test design with power analysis, and competitive positioning against Frontify, Bynder, and Canva's own Brand Assist. React + TypeScript + Hono backend + Neon PostgreSQL.
Designed and shipped the Consistency Agent, an AI system that catches when your marketing campaign says different things across different formats. Brand Kit makes sure your designs look the same. The Consistency Agent makes sure they say the same thing. It scans every asset in a campaign using a three-tier taxonomy: Must-Match (CTAs, dates, prices, URLs), Should-Match (tone, hero image), and Format-Dependent (layout, text length). One-click Sync All propagates the correct version everywhere. Built a full working prototype, campaign dashboard with 8 asset types, real Claude API integration with streaming, live drift detection in the canvas editor, brief upload with canonical value extraction, and a 9-step guided onboarding tour. AI architecture uses semantic role classification, sentence embeddings (MiniLM), perceptual hashing, and rule-based propagation. Spec included 24-week ship plan, A/B test design with power analysis, and competitive positioning against Frontify, Bynder, and Canva's own Brand Assist. React + TypeScript + Hono backend + Neon PostgreSQL.
Prototype for Lora (Backed By alleycorp)
Prototype for Lora (Backed By alleycorp)
Built a growth prototype for Lora, a consumer identity app that routes users into personalized experiences based on who they are. The core challenge: Lora had deep engagement within their community (6,000+ Discord members in 3 months, 50%+ newsletter open rates) but needed new surfaces to drive frequency and top of funnel growth. I built a zero-friction viral loop, users enter their phone number, get a personalized identity reading they actually want to share. No signup, no download, just instant value that spreads. Beyond acquisition, I pitched four product ideas to deepen retention: monthly engagement hooks tied to recurring identity cycles, a shareable content engine that turns every user into a distribution channel, expanding Match Lab's routing precision from 3 dimensions to 12 (going from surface-level matching to real depth), and ambient widgets that give users a reason to come back daily instead of weekly. The thesis was that Lora's moat isn't the data, it's the trust users place in identity based routing. More precision means more trust, more trust means more engagement, more engagement means the matching gets better. Shipped interactive pitch decks personalized for two team members and a working Python prototype demonstrating the viral loop. Deployed on Vercel.
Built a growth prototype for Lora, a consumer identity app that routes users into personalized experiences based on who they are. The core challenge: Lora had deep engagement within their community (6,000+ Discord members in 3 months, 50%+ newsletter open rates) but needed new surfaces to drive frequency and top of funnel growth. I built a zero-friction viral loop, users enter their phone number, get a personalized identity reading they actually want to share. No signup, no download, just instant value that spreads. Beyond acquisition, I pitched four product ideas to deepen retention: monthly engagement hooks tied to recurring identity cycles, a shareable content engine that turns every user into a distribution channel, expanding Match Lab's routing precision from 3 dimensions to 12 (going from surface-level matching to real depth), and ambient widgets that give users a reason to come back daily instead of weekly. The thesis was that Lora's moat isn't the data, it's the trust users place in identity based routing. More precision means more trust, more trust means more engagement, more engagement means the matching gets better. Shipped interactive pitch decks personalized for two team members and a working Python prototype demonstrating the viral loop. Deployed on Vercel.
Prototype for Eulerity
Prototype for Eulerity
Built a franchise performance intelligence dashboard for Eulerity's marketing automation platform using real Kumon data, 55 NYC metro locations with live performance metrics. Network overview with interactive map and composite health scores. Location drill-down with sparkline trend lines, spend breakdowns, and AI-generated plain-language insights per location. Benchmarks table for cross-location comparison so franchise owners can see how they stack up. Authored "The Adoption Stack", a four-bet product strategy: Location Scorecard (single 0-100 health score to convert passive franchisees into weekly active users), AI Campaign Copilot (natural language interface, "spend more on weekends" to lower the barrier for non-technical franchise owners), Territory Intelligence (geospatial analysis showing overlap, saturation, and competitive density), and AI Search Visibility (tracking how locations appear in ChatGPT, Perplexity, and Google AI Overviews). Competitive analysis against SOCi, Scorpion, and Yext. Next.js + Vercel.
Built a franchise performance intelligence dashboard for Eulerity's marketing automation platform using real Kumon data, 55 NYC metro locations with live performance metrics. Network overview with interactive map and composite health scores. Location drill-down with sparkline trend lines, spend breakdowns, and AI-generated plain-language insights per location. Benchmarks table for cross-location comparison so franchise owners can see how they stack up. Authored "The Adoption Stack", a four-bet product strategy: Location Scorecard (single 0-100 health score to convert passive franchisees into weekly active users), AI Campaign Copilot (natural language interface, "spend more on weekends" to lower the barrier for non-technical franchise owners), Territory Intelligence (geospatial analysis showing overlap, saturation, and competitive density), and AI Search Visibility (tracking how locations appear in ChatGPT, Perplexity, and Google AI Overviews). Competitive analysis against SOCi, Scorpion, and Yext. Next.js + Vercel.
Three Axis
Three Axis
Market research workspace for time price and cycle views
Tech Stack: React, Next, Python, Charting libraries, Firebase
Our terminal is a sophisticated data analysis platform combining real-time market data with advanced cyclic pattern recognition. It's engineered for traders and analysts who require precision timing and a deeper understanding of market structures.THREE AXIS was founded on the principle that conventional financial analysis is constrained by a limited set of dimensions. Our research is focused on identifying and quantifying the persistent, underlying forces that drive market behavior. We employ mathematical and statistical methods to design and execute our analytical models.
Our approach is rooted in a culture of rigorous, independent thinking. We believe in a systematic, data-first methodology, removing emotion and cognitive bias from the analytical process.
Market research workspace for time price and cycle views
Tech Stack: React, Next, Python, Charting libraries, Firebase
Our terminal is a sophisticated data analysis platform combining real-time market data with advanced cyclic pattern recognition. It's engineered for traders and analysts who require precision timing and a deeper understanding of market structures.THREE AXIS was founded on the principle that conventional financial analysis is constrained by a limited set of dimensions. Our research is focused on identifying and quantifying the persistent, underlying forces that drive market behavior. We employ mathematical and statistical methods to design and execute our analytical models.
Our approach is rooted in a culture of rigorous, independent thinking. We believe in a systematic, data-first methodology, removing emotion and cognitive bias from the analytical process.
Cycle X
Financial Markets Future Price Projection
Cycle X
Financial Markets Future Price Projection
Tech Stack: Firebase, React, Framer, JavaScript,Python, Dash, and Plotly
Offers a glimpse into potential future price levels—all with an interactive, easy-to-use interface. I built the My Proprietary Cycle Analyzer + Projection dashboard to do just that, I created a dynamic web app that detects key pivot points in historical data and then lets you project future price targets simply by clicking on these points.
Tech Stack: Firebase, React, Framer, JavaScript,Python, Dash, and Plotly
Offers a glimpse into potential future price levels—all with an interactive, easy-to-use interface. I built the My Proprietary Cycle Analyzer + Projection dashboard to do just that, I created a dynamic web app that detects key pivot points in historical data and then lets you project future price targets simply by clicking on these points.
Flengg
Flengg
Social habit challenges with close friends
Tech Stack: React, TypeScript, Firebase, Vercel
Flengg is a small group challenge app where friends set weekly goals, check in, and keep score in one place. The core flow is simple join a challenge, log progress, see crowns and streaks update in real time. The focus is on making habit tracking feel like a game instead of another productivity tool.
Social habit challenges with close friends
Tech Stack: React, TypeScript, Firebase, Vercel
Flengg is a small group challenge app where friends set weekly goals, check in, and keep score in one place. The core flow is simple join a challenge, log progress, see crowns and streaks update in real time. The focus is on making habit tracking feel like a game instead of another productivity tool.
Brain
Brain
Binaural sound studio for focus and deep work
Tech Stack: React, Web Audio API, Firebase
Brain is a browser based audio app where users pick a goal focus relax or sleep and get a matching binaural session instantly. It layers tones and ambient loops with the Web Audio API and wraps them in a clean neon style interface. Simple controls for timers and smooth fade in and fade out help Brain fit into real study or work blocks without friction.
Binaural sound studio for focus and deep work
Tech Stack: React, Web Audio API, Firebase
Brain is a browser based audio app where users pick a goal focus relax or sleep and get a matching binaural session instantly. It layers tones and ambient loops with the Web Audio API and wraps them in a clean neon style interface. Simple controls for timers and smooth fade in and fade out help Brain fit into real study or work blocks without friction.
RAG Chatbot in AWS Bedrock
RAG Chatbot in AWS Bedrock
I created a personalized chatbot using Amazon Bedrock—one that can answer questions based on my own documents!
The best part? My chatbot only responds with verified information from my documents—no random AI hallucinations!
Here’s what I did:
Set up a Knowledge Base to store and process documents in S3
Used vector search in OpenSearch Serverless to retrieve relevant info
Integrated Llama AI models for natural, human-like responses
Fine-tuned performance to improve accuracy
This project was a great hands-on way to explore AWS services, RAG (Retrieval-Augmented Generation), and AI-powered search. Excited to keep learning and building!
I created a personalized chatbot using Amazon Bedrock—one that can answer questions based on my own documents!
The best part? My chatbot only responds with verified information from my documents—no random AI hallucinations!
Here’s what I did:
Set up a Knowledge Base to store and process documents in S3
Used vector search in OpenSearch Serverless to retrieve relevant info
Integrated Llama AI models for natural, human-like responses
Fine-tuned performance to improve accuracy
This project was a great hands-on way to explore AWS services, RAG (Retrieval-Augmented Generation), and AI-powered search. Excited to keep learning and building!
Sales Performance Dashboard using Tableau
Sales Performance Dashboard using Tableau
The Sales Performance Dashboard project involved creating two comprehensive Tableau dashboards designed specifically for stakeholders, including sales managers and executives, to effectively analyze sales performance, identify trends, and understand customer behavior. Utilizing real-world data from sales transactions, customer demographics, locations, and products, these dashboards provide actionable insights into yearly and monthly sales metrics.
Dashboard Features and Highlights
1. KPI Overview
2. Monthly Sales Trends
3. Product Subcategory Comparison
4. Weekly Trends for Sales & Profit
The Sales Performance Dashboard project involved creating two comprehensive Tableau dashboards designed specifically for stakeholders, including sales managers and executives, to effectively analyze sales performance, identify trends, and understand customer behavior. Utilizing real-world data from sales transactions, customer demographics, locations, and products, these dashboards provide actionable insights into yearly and monthly sales metrics.
Dashboard Features and Highlights
1. KPI Overview
2. Monthly Sales Trends
3. Product Subcategory Comparison
4. Weekly Trends for Sales & Profit
Ai Cloud Text Expander
Ai Cloud Text Expander
Developed a Cloud Text Expander application that enhances productivity by providing auto-complete, dynamic shortcuts, and templates for frequently used text. Utilized Python and cloud services to ensure real-time synchronization across devices. Designed an intuitive interface and implemented user-specific data security measures.
Developed a Cloud Text Expander application that enhances productivity by providing auto-complete, dynamic shortcuts, and templates for frequently used text. Utilized Python and cloud services to ensure real-time synchronization across devices. Designed an intuitive interface and implemented user-specific data security measures.
EV Charging Management System Product
EV Charging Management System Product
Developed a robust software solution using C# to manage electric vehicle (EV) charging station schedules efficiently. Leveraged priority-based scheduling algorithms to reduce average wait times by 35%. Integrated an SQL database for managing operational metadata, enhancing reporting and analysis capabilities by 40%.
Designed predictive models for return on investment (ROI) analysis, aligning projections with actual business outcomes.
The project focused on building an intuitive user interface and ensuring compatibility with smart grid systems for real-time demand management.
Skills: C#, SQL, Data Modeling, ROI Optimization, Software Engineering.
Developed a robust software solution using C# to manage electric vehicle (EV) charging station schedules efficiently. Leveraged priority-based scheduling algorithms to reduce average wait times by 35%. Integrated an SQL database for managing operational metadata, enhancing reporting and analysis capabilities by 40%.
Designed predictive models for return on investment (ROI) analysis, aligning projections with actual business outcomes.
The project focused on building an intuitive user interface and ensuring compatibility with smart grid systems for real-time demand management.
Skills: C#, SQL, Data Modeling, ROI Optimization, Software Engineering.
Git
w
Predictive Analytics and Database for Health Insurance
Predictive Analytics and Database for Health Insurance
Built a machine learning pipeline to predict insurance costs with models like Linear Regression and Random Forest, achieving a test R² score of 0.86.
Designed and implemented a SQL database to manage over 1,300 patient records, improving data retrieval efficiency by 20% and
ensuring data integrity.
Conducted data analysis to uncover key factors influencing costs, enabling actionable business insights.
Enhanced marketing effectiveness by 25% through segmentation and high-cost group predictions.
Built a machine learning pipeline to predict insurance costs with models like Linear Regression and Random Forest, achieving a test R² score of 0.86.
Designed and implemented a SQL database to manage over 1,300 patient records, improving data retrieval efficiency by 20% and
ensuring data integrity.
Conducted data analysis to uncover key factors influencing costs, enabling actionable business insights.
Enhanced marketing effectiveness by 25% through segmentation and high-cost group predictions.
Git
w
Moody
Self Health - Voice/picture journaling web app
Moody
Self Health - Voice/picture journaling web app
Tech Stack: Firebase, React, Framer, JavaScript, google cloud run.
Moody is a lightweight, voice-based journaling tool designed to help users track their moods with minimal effort. Users simply snap a photo, choose an emoji-based vibe, and leave a voice note. Past vibes can be viewed through a visual calendar and sorted by mood. Designed for emotional reflection without the burden of traditional journaling.
Tech Stack: Firebase, React, Framer, JavaScript, google cloud run.
Moody is a lightweight, voice-based journaling tool designed to help users track their moods with minimal effort. Users simply snap a photo, choose an emoji-based vibe, and leave a voice note. Past vibes can be viewed through a visual calendar and sorted by mood. Designed for emotional reflection without the burden of traditional journaling.
Dripper Design and Manufacturing ProjectDripper Design and Manufacturing Project
Dripper Design and Manufacturing ProjectDripper Design and Manufacturing Project
Dripper Design and Manufacturing Project
Description:
Successfully led the end-to-end design and manufacturing lifecycle of a high-quality micro-irrigation dripper, tailored to the specific needs of modern agriculture. Directed cross-functional teams, including engineers and suppliers, to deliver the project on schedule and within budget. Applied product management best practices to ensure the dripper design optimized water distribution efficiency, meeting stringent quality and performance standards. Leveraged Primavera P6 to meticulously coordinate project timelines, identify bottlenecks, and execute tasks seamlessly. Collaborated with agricultural experts to incorporate user feedback, enhancing the product's usability and market relevance.
Key Achievements:
Delivered a cost-effective and efficient irrigation solution, boosting crop yield potential.
Streamlined the project workflow, achieving a 15% reduction in production time.
Skills: Project Management, Product Design, Primavera P6, Team Coordination, Agricultural
Dripper Design and Manufacturing Project
Description:
Successfully led the end-to-end design and manufacturing lifecycle of a high-quality micro-irrigation dripper, tailored to the specific needs of modern agriculture. Directed cross-functional teams, including engineers and suppliers, to deliver the project on schedule and within budget. Applied product management best practices to ensure the dripper design optimized water distribution efficiency, meeting stringent quality and performance standards. Leveraged Primavera P6 to meticulously coordinate project timelines, identify bottlenecks, and execute tasks seamlessly. Collaborated with agricultural experts to incorporate user feedback, enhancing the product's usability and market relevance.
Key Achievements:
Delivered a cost-effective and efficient irrigation solution, boosting crop yield potential.
Streamlined the project workflow, achieving a 15% reduction in production time.
Skills: Project Management, Product Design, Primavera P6, Team Coordination, Agricultural
Keurig Dr. Pepper's Alcoholic Beverage Project
Keurig Dr. Pepper's Alcoholic Beverage Project
Managed Keurig Dr. Pepper's innovative alcoholic beverage project, blending expertise in project management, marketing strategy, and consumer analysis. Conducted comprehensive market research to understand consumer preferences and industry trends, which informed critical strategic decisions. Applied Primavera P6 to manage intricate project timelines, ensuring smooth collaboration across marketing, production, and distribution teams. Designed and implemented a marketing plan that aligned with consumer trends, leveraging data-driven insights to target key demographics and maximize brand visibility.
Key Achievements:
Successfully launched the product within the targeted timeline, capturing significant market share in the first quarter.
Established strategic partnerships, enhancing the product's market penetration and brand equity.
Managed Keurig Dr. Pepper's innovative alcoholic beverage project, blending expertise in project management, marketing strategy, and consumer analysis. Conducted comprehensive market research to understand consumer preferences and industry trends, which informed critical strategic decisions. Applied Primavera P6 to manage intricate project timelines, ensuring smooth collaboration across marketing, production, and distribution teams. Designed and implemented a marketing plan that aligned with consumer trends, leveraging data-driven insights to target key demographics and maximize brand visibility.
Key Achievements:
Successfully launched the product within the targeted timeline, capturing significant market share in the first quarter.
Established strategic partnerships, enhancing the product's market penetration and brand equity.
Simulation of Dual Three Phase Induction Motor Control Using Five-leg Inverter
Conducted an advanced simulation study to control dual three-phase induction motors using a shared five-leg inverter. The project aimed to evaluate motor performance under variable speed demands and demonstrate precise and independent regulation of both motors. Leveraged MATLAB/Simulink to design, program, and analyze control strategies, including Pulse Width Modulation (PWM) generation and gate driver logic. Explored and implemented a variety of modulation methods, such as sinusoidal PWM, two-arm modulation, and double zero sequence modulation, to achieve optimized motor performance.
Key Contributions:
Successfully simulated and analyzed independent speed control of two induction motors powered by a single five-leg inverter.
Optimized control algorithms to minimize harmonic distortion while maintaining power efficiency.
Demonstrated the efficacy of advanced modulation techniques for motor control in real-world industrial scenarios.
Key Achievements:
Validated the feasibility of using a five-leg inverter for dual motor control, reducing hardware costs compared to traditional six-leg inverters.
Showcased the project findings through simulations, presenting data-driven insights on performance metrics such as torque ripple, harmonic distortion, and power utilization.
Skills: Motor Control, MATLAB/Simulink, PWM Generation, Inverter Modulation Techniques, Simulation Analysis.