Six years of commits, in the open and behind closed doors
A summary of my GitHub activity — what I have built, how my stack evolved and where the work went. Client and private repositories are described in anonymised form: stack and scope only, never names or code.
63 public · 27 private and client-owned
Commits, PRs and reviews since 2021
First commit pushed in 2020
Plus Python, Solidity and C++
Where the work went, year by year
The darker portion of each bar is private and client-owned work. The shift from public learning to private delivery is the clearest signal of how my career moved.
Fundamentals: React, Express and MongoDB side projects
First full-time role — production backends and e-commerce work
Microservices, NestJS, gRPC and Docker deep-dives in the open
Freelance and contract delivery — scraping, admin panels, trading systems
Busiest year: AI agents, Web3 platforms and client product builds
Peaklyft — banking application maintenance and internal platform work
What I actually write
Nine builds worth opening
Distributed job engine
A NestJS microservices job engine that fans work out over Apache Pulsar, with GraphQL for clients and gRPC between services. Built to explore back-pressure and retry semantics under load.
Trading / exchange simulator
PrivateAn order-matching stock simulator handling concurrent order flow. Redis and BullMQ queue the matching pipeline so trades settle deterministically while the UI streams live book updates.
MERN Space — auth & catalog microservices
Production-grade authentication and catalog services: refresh-token rotation, RBAC, migrations, structured logging and a full test suite — the reference architecture I reuse on client backends.
SERP scraping engine
A production scraping service that collects and normalises search-engine result data at scale, with an async worker pool, proxy rotation and a separate dashboard frontend for the client.
AI agent platform
A SaaS AI agent platform on Next.js 15 with real-time meeting agents, auth and subscription billing wired end to end — the open counterpart to the voice-agent work I ship for clients.
Enterprise sales-practice voice agent
PrivateA LiveKit-based voice agent enterprises use to rehearse sales calls: real-time speech in and out, scenario scripting, and automated scoring of each conversation against a rubric.
Solana launchpad & DeFi work
PrivateToken launchpad, presale and bonding-curve integrations on Solana for a Web3 studio, including wallet flows, transaction retry handling and RPC failover under launch-day traffic.
Workflow automation platform
PrivateA drag-and-drop automation builder written from scratch: a visual graph editor on the frontend and a durable execution engine that runs, retries and audits each node server-side.
Real-time chat & socket scaling
Horizontally scaled WebSocket infrastructure — Redis pub/sub adapters, sticky sessions and load tests — alongside a monolithic social backend using Express, Redis and BullMQ.
How the stack evolved
Fundamentals in the open
Twenty repositories in one year, mostly React and vanilla JavaScript builds — typing apps, expense managers, e-commerce clones — learning the DOM, state and REST the hard way.
Backend depth and first production code
Moved from tutorials to shipping: authentication systems, file uploads, API documentation with Swagger, and my first full-time backend work behind private repositories.
Distributed systems and DevOps
The year of breaking monoliths apart — an airline booking system split into auth, flights, booking and reminder microservices, plus gRPC, GraphQL, Neo4j, Docker and NGINX reverse proxies.
TypeScript everywhere, contract delivery
TypeScript became the default. Work shifted to client contracts — scraping infrastructure, multi-tenant admin panels, Azure pipelines and a trading simulator built around queues.
AI agents and Web3 at production scale
LiveKit voice agents, LangChain and vector retrieval, Solana launchpad contracts, and a workflow automation engine — with most of it now behind private, client-owned repositories.