DNS Server
A minimal DNS server and client in Node.js over UDP sockets.
I'm Sai Teja - a backend and AI engineer with 3+ years building distributed systems, enterprise platforms, and AI-powered products. I care about writing clean, efficient code and understanding the system end-to-end, not just the layer I own.
Currently at Dhan AI, I've shipped across customer portals, internal operations tooling, and a large-scale multi-tenant CXP platform. I'm drawn to hard problems - event-driven architectures, microservice coordination, and systems that need to be resilient at scale.
I'm equally deep into AI engineering - building with LLMs, RAG pipelines, and autonomous agents. I've worked with retrieval-augmented systems, prompt-engineered production workflows, and I actively track the frontier of open-source AI. For me, the most exciting work happens where backend systems meet AI - building the infrastructure that makes models useful in the real world.
Outside of work, I'm solving problems on LeetCode and posting about tech on Instagram - because explaining something well is how I know I actually get it.
Technologies I work with, from the services I ship at work to what I build on the side.
A customer-facing platform enabling users to view and manage active services in real time. Architected as a middleware microservice layer, bridging the frontend with multiple external service providers - handling auth, session management, and data orchestration.
Internal tooling for support agents to manage services and handle trouble tickets end-to-end. Built a proactive network diagnostics engine that detects and resolves infrastructure issues before customers are impacted - reducing dispatches and cutting resolution costs.
Modernized a large-scale customer experience platform with 15+ microservices powering onboarding, billing, payments, and order lifecycle. Delivered 30% improvement in API response times. Integrated Salesforce, Dynamics 365, and Firebase RBAC.
Things I build outside work to understand how systems really work, most with a write-up on the blog.
A working Python RAG pipeline: chunking, embeddings, ChromaDB, hybrid search, cited answers and evaluation.
An Express API that estimates unique users at scale with HyperLogLog, in about 12 KB of memory.
An HTTP/1.1 server on raw TCP sockets in Node.js: routing, file serving, uploads and gzip.
10 projects across 5 languages
A minimal DNS server and client in Node.js over UDP sockets.
A REST API for tasks with Spring Boot, Spring Data JPA and MySQL.
Employees, locations and projects over Express and TypeORM on SQL Server, with migrations and validation.
Scaling Socket.IO across server instances with the MongoDB adapter and emitter, tested with multiple clients.
A GraphQL layer on Apollo Server with TypeORM entities for books, characters and houses.
A monthly habit grid with streaks and a live progress chart.
Building distributed backend systems and AI-powered features serving enterprise and consumer clients. Designing LLM-integrated workflows and scalable API platforms across customer portals, operations tooling, and a multi-tenant CXP platform.
Python scripting and cybersecurity tooling. Applied automation to security workflows and vulnerability assessments.
Intensive Data Structures & Algorithms training. Built strong foundations in problem-solving and algorithmic thinking that underpin every system I design today.
Hands-on experience in threat analysis, security tooling, and foundational cybersecurity certification.
Notes on backend systems, distributed architecture and AI engineering.
A beginner-friendly walkthrough of RAG: cleaning, chunking, embedding and storing data, then multi-query, hybrid search, fusion, reranking and cited answers.
A guide to Spec-Driven Development with GitHub Spec Kit: the workflow, setup for new and existing repos, every speckit command, and agent integrations.
How HyperLogLog counts billions of unique users in 12 KB, why it beats sets and Bloom filters for cardinality, and when you should not use it.
Where an agent's tokens actually go, one tool for each of the three leaks, and what an independent benchmark measured versus what each tool advertises.
Open for backend engineering, AI/ML engineering, and full-stack AI roles. Whether it's building production LLM systems, distributed architectures, or something at the intersection - my inbox is always open.