Open to collaboration
Building
Privacy-First
Software Products
Founder & Senior Software Engineer. I craft high-performance, AI-driven SaaS solutions
around a single principle: process sensitive data locally to reduce liability and maximize speed.
What I’ve built
Products
ChatNexus — AI assistant builder with built-in scheduling
Build smart AI assistants that answer questions, book appointments, and act on your behalf.
Train the assistant using a knowledge base (documents, pasted text, or URLs), stream
responses in real time, and embed the widget on your website.
Key highlights
- Knowledge base ingestion and RAG using PostgreSQL + pgvector
- Multi-provider LLM routing — swap providers without rewriting core logic
- Real-time streaming chat for fast, human-like responses
- Built-in scheduling and booking flows
CleanCrumbs — Privacy-first CRM lead sanitizer
Sanitize and validate CRM lead exports (CSV or XLSX) entirely inside the browser.
Your file data does not get uploaded to a server — ever.
Key highlights
- Zero-server processing — data stays entirely on the device
- Cleans names, emails, phones, and addresses for CRM import readiness
- Clear privacy positioning and verification guidance
ImageCraft Hub — Privacy-first image editor (PWA, works offline)
A browser-based image editing platform with professional tools that run entirely in the browser.
Built as a Progressive Web App — installable and fully usable offline after the first visit.
All image processing happens locally; no uploads, no server, no privacy risk.
Key highlights
- 100% client-side image processing using HTML5 Canvas + WebAssembly (WASM)
- Installable PWA with full offline support after first visit
How I build
Case Studies
A look inside the architectural decisions, engineering trade-offs, and design principles behind each product.
ChatNexus — Architecture & Trade-offs
Problem
Businesses need an AI assistant trained on their own content that can answer questions and handle bookings — without building complex infrastructure from scratch.
Solution
A multi-tenant platform with knowledge base ingestion, vector search via pgvector, multi-provider LLM routing, and real-time streaming responses — shipped as a white-label widget.
CleanCrumbs — Privacy-first Approach
Problem
CRM lead lists contain sensitive personal data. Uploading CSV files to third-party servers for cleaning creates GDPR exposure and unnecessary liability.
Solution
A browser-only sanitizer where files stay local. All parsing, validation, and field cleaning happens on-device using vanilla JS — zero server round trips, zero data exposure.
Technical writing
Featured Writing
Deep dives into system design, distributed systems, and building production-grade software.
System Design — Video Streaming Platform
A system design breakdown for a YouTube-style video platform, covering capacity estimation, upload pipelines, CDN architecture, and playback strategies.
Understanding Eventual Consistency in Distributed Systems
A practical explanation of eventual consistency, the CAP theorem, why it exists, and the real-world trade-offs it introduces for distributed data stores.
I Built a Custom AI Assistant Platform
Architectural decisions and lessons learned while building ChatNexus — multi-provider LLM routing, pgvector RAG, real-time streaming, and scheduling integration.
Background
About Me
Hi, I’m Chisty. I build scalable systems, design clean microservices, and solve
real-world backend challenges — with a particular obsession for privacy-first engineering.
My engineering philosophy: process sensitive data locally wherever possible, minimise
server-side risk, and deliver fast, reliable experiences through deliberate architectural
trade-offs.
Products shipped
Client-side processing in privacy tools
Server uploads in privacy-first products
Trade-offs carefully considered
Let’s talk
Get in Touch
Open to collaboration, consulting, and interesting product conversations.
Reach out through any of the channels below.