W
selected work

22+ Products

Shipped to Production

From AI platforms generating revenue to fintech apps processing real transactions. These aren't mockups. They're live products solving real business problems.

AI SaaS Platform

Reelzila

reelzila.studio

AI video platform with a creator marketplace that generates 40-70% profit margins.

6
AI Models
40-70%
Profit Margin
50+
API Endpoints
8 weeks
Time to Market
Reelzila — HomepageReelzila — AI Models

The Challenge

Content creators needed access to cutting-edge AI video generation without paying enterprise prices. Existing tools were either too expensive, limited to one model, or had no way for creators to monetize their generations.

Our Solution

We built a complete AI video platform with 6 generation models (Google VEO, OpenAI Sora, Kling, LTX, Hailuo, Nano Banana), a credit-based pricing system, and a peer-to-peer marketplace where creators sell their AI-generated content to buyers.

Built With

Next.js 16React 19FastAPIFirebaseFal.aiVercel

What We Delivered

01

Multi-model AI generation engine with per-model pricing

02

Creator marketplace with seller dashboards and payout system

03

Dynamic credit system ($0.05/credit) with transparent pricing

04

Admin panel with user management, seller verification, and analytics

05

PayTrust payment gateway with webhook verification

06

Security audit: 0 critical vulnerabilities

AI Sales Intelligence

Magnet

Autonomous lead generation that discovers, qualifies, and contacts prospects while you sleep.

9+
Lead Sources
12
Service Verticals
2x
Daily Schedule
4-step
Email Sequence
Magnet — PipelineMagnet — Dashboard

The Challenge

Sales teams waste 60-70% of their time on manual prospecting — browsing job boards, scanning forums, copy-pasting emails. By the time they reach a lead, competitors have already made contact. The process doesn't scale.

Our Solution

We built an autonomous lead generation engine that scrapes 9+ sources (Upwork, Reddit, Hacker News, Indeed, RemoteOK, and more), classifies signals with a multi-layer confidence scorer, and generates personalized outreach sequences via Claude AI — all running hands-free on a daily schedule.

Built With

FastAPIPlaywrightClaude AIResendWebSocketPython

What We Delivered

01

9-source scraper engine: Upwork CDP, Reddit, HN, Indeed, RemoteOK, Freelancer, and more

02

Multi-layer confidence scoring (source + keywords + budget + urgency + signal type)

03

Claude-powered personalized email generation with case study matching

04

4-email automated sequence: initial → day 3 → day 7 → day 14 follow-ups

05

Self-improving classifier via training loop with AI feedback

06

macOS daemon: runs at 6am & 3pm daily with catch-up logic

B2B Intelligence Platform

Sonar

B2B buying signal radar. Detects companies ready to buy before your competitors do.

10+
Signal Sources
6
Signal Types
75+
API Endpoints
5
Enrichment Steps
Sonar — Companies

The Challenge

B2B sales teams rely on LinkedIn and outdated databases to find leads. They miss buying signals — funding rounds, leadership changes, tech migrations, expansion announcements — that indicate a company is actively ready to purchase. By the time they find out, the deal is already in motion.

Our Solution

We built a 5-stage intelligence pipeline: Discover (10+ sources), Classify (6 signal types), Connect (entity resolution), Score (AI-powered lead ranking), and Act (automated outreach with tracking). A Clay-style enrichment waterfall extracts contacts and emails automatically.

Built With

FastAPIPlaywrightClaude AIResendaiohttpPython

What We Delivered

01

10-source monitoring: Google News, Indeed, Crunchbase, HN, Product Hunt, GitHub, LinkedIn, G2

02

5-stage pipeline: Discover → Classify → Connect → Score → Act

03

Claude-powered lead scoring with rule-based fallback (24h cache, 40% cost reduction)

04

Clay-style enrichment waterfall: website → team → contacts → emails → scoring

05

Multi-step outreach sequences with Resend email tracking (opens, clicks, replies)

06

Real-time WebSocket dashboard with live signal feed

Digital Archive Platform

Heritage Vault

AI-powered authenticity verification for cultural preservation. Trusted by institutions.

5
Verification Stages
3
Languages
14
Cultural Categories
50+
API Endpoints
Heritage Vault — HomepageHeritage Vault — AI Verify

The Challenge

Cultural institutions needed a trusted repository to preserve historical media while combating deepfakes and AI-generated forgeries. Existing platforms had no way to cryptographically verify that content was authentic and untampered.

Our Solution

We built a digital archive platform with multi-stage verification: SHA-256 hashing, ECDSA digital signatures, perceptual fingerprinting, and AI-generated content detection. Role-based access for archivists, curators, and researchers across 3 languages.

Built With

Next.js 16PostgreSQLPrismaCloudflare R2NextAuthSharp

What We Delivered

01

AI detection engine: identifies Stable Diffusion, DALL-E, Midjourney, RunwayML content

02

Cryptographic verification: SHA-256 + ECDSA-P256 digital signatures

03

Perceptual hashing with deduplication (Hamming distance similarity search)

04

Complete audit trail for institutional compliance

05

Role-based access: Admin, Curator, Contributor, Viewer

06

Trilingual: English, French, Malagasy with 22 region mappings

Your Project

Ready To Build
Something
Like This?

Every product above started with a conversation. Tell us what you're building and we'll tell you exactly what it takes.