AI that ships.
RAG pipelines, AI agents, embeddings, and local inference. Generative AI integrated into real production workflows.
Full-stack software engineer · Bangalore, India
Engineering
the next.↗
I’m Shubhankar Kumar Singh.
I build AI-native products, real-time experiences, and
distributed systems that hold up beyond the demo.
Generative AI
Distributed systems
End-to-end ownership
From database to interface.
Scroll to explore ↓
5 yrs
Full-stack engineering experience
10K+
Concurrent users · AI chat platform
100K+
Notification events processed daily
99.9%
Production uptime achieved
Systems that scale.
Event-driven services, asynchronous processing, distributed rate limiting, and high-concurrency real-time platforms.
Ownership that lasts.
Secure APIs, responsive interfaces, cloud deployments, and observability. Built with reliability in mind from the start.
01 / Selected open source
Less theory.
More shipped.
Focused tools for difficult engineering problems. Built for concurrency, practical AI, and reuse.
Redis-Lua-Rate-Limiter
01A distributed, atomic token-bucket rate limiter for Node.js. Redis Lua scripts keep enforcement consistent across horizontally scaled instances—even under heavy concurrency.
≈ 4,500 requests/sec · 1,000 concurrent users · 200+ active NPM users
Inside the engineering
- Atomic token-bucket operations implemented with Redis Lua scripts.
- Strict rate-limit enforcement across horizontally scaled instances.
- No burst leaks or race conditions under the reported concurrency test.
- Published as a reusable NPM package for Node.js services.
EmbedServe-ORT
02An OpenAI-compatible embeddings server powered by Node.js and ONNX Runtime. Run transformer models locally on CPUs, without external inference APIs or network round trips.
CPU-only inference · Fully offline · No external inference API costs
Inside the engineering
- OpenAI-compatible interface for integration with existing AI workflows.
- Attention-aware mean pooling and L2 normalization.
- Embeddings optimized for Pinecone and MongoDB Atlas Vector Search.
- Local inference designed for microservices and cloud-native deployments.
02 / The engineering toolkit
Different tools.
One connected stack.
From the interface to the inference pipeline. Choose a category, then select a technology to explore.
Technology explorer
05 technologies
Built around
System design
Event-driven architecture
Horizontal scaling
Fault tolerance
API security
Observability
View the complete toolkit as text
Languages & frontend
JavaScript, TypeScript, Python, SQL, Java, React.js, Redux Toolkit, Tailwind CSS, responsive UI, state management, frontend–backend integration.
Backend & architecture
Node.js, Express.js, FastAPI, RESTful APIs, GraphQL, WebSockets, SSE, microservices, distributed systems, asynchronous processing, high availability.
AI & inference
Generative AI, LLM integration, AI agents, RAG, semantic search, embeddings, LangChain, OpenAI, Azure OpenAI, Pinecone, MongoDB Atlas Vector Search, ONNX Runtime.
Databases & messaging
PostgreSQL, MongoDB, MySQL, Redis, RabbitMQ, BullMQ, data modeling, schema design, query optimization, retries, delivery guarantees, distributed rate limiting.
Cloud & observability
AWS EC2, S3, Lambda, SES, Docker, CI/CD, Firebase, Cloudflare Workers, Grafana, Loki, Sentry, Git, production monitoring, incident troubleshooting, RCA.
Security & performance
RBAC, authentication, authorization, input validation, secure API design, reCAPTCHA Enterprise, concurrency tuning, latency optimization, throughput optimization, SDK development.
03 / Beyond the commit
Performance you
can put a number on.
Selected outcomes from production engineering. Different systems. The same focus on measurable improvement.
Playhouse Media / Voice AI
40% ↓
Lower interaction latency
Deepgram streaming speech-to-text, asynchronous processing, and Node.js concurrency optimization for a real-time voice pipeline.
Troopr Labs / Real-time AI
50ms
Median chat latency
A WebSocket and LangChain AI chat platform supporting 10K+ concurrent users through high-concurrency distributed processing.
Playhouse Media / Observability
80% ↓
Less time resolving issues
Centralized logging, metrics, and production monitoring with Grafana and Loki to support troubleshooting and root-cause analysis.
04 / The journey
Real products.
Real ownership.
From cloud infrastructure to enterprise AI. Five years of building, optimizing, debugging, and shipping across the entire application stack.
Connect on LinkedIn ↗Apr 2025 — Jul 2025 · Remote, US
Playhouse Media
Senior Software Development Engineer · Independent Contractor, Full-time
- Shipped scalable React and Node.js features for an internal operations platform: RBAC, authentication, secure APIs, Stripe billing, and administrative workflows.
- Engineered a Deepgram streaming voice AI pipeline with asynchronous processing and concurrency tuning, reducing interaction latency by 40%.
- Built BullMQ and Redis automation microservices for scheduled campaigns, background jobs, retries, and reliable task execution.
- Implemented Grafana and Loki observability, reducing issue resolution time by 80% and improving production troubleshooting.
- Owned delivery across interfaces, backend services, APIs, data workflows, and infrastructure.
Nov 2023 — Dec 2024 · Bangalore
Troopr Labs
Software Development Engineer · Consultant, Full-time
- Built Enjo, an AI-powered ticket resolution agent with Jira and Salesforce integrations for case summarization, contextual answers, and knowledge generation.
- Implemented a RAG pipeline using OpenAI embeddings and MongoDB Vector Search to retrieve context from historical tickets, enterprise knowledge bases, and unstructured support data.
- Delivered WebSocket and LangChain AI chat supporting 10K+ concurrent users at 50ms median latency.
- Designed a unified Slack and web inbox for multi-channel conversations, combining real-time events with backend workflows.
- Improved AI request throughput by 40% using Node.js Cluster and Worker Threads.
Feb 2022 — Aug 2023 · Noida
Zeeve
Software Engineer · Full-time
- Designed and owned an event-driven notification microservice using Node.js, RabbitMQ, FCM, and PostgreSQL, processing 100K+ events per day.
- Built a reusable notification SDK standardizing event publishing, retries, asynchronous processing, and delivery guarantees.
- Strengthened critical APIs with reCAPTCHA Enterprise to reduce automated bot and fraudulent traffic.
- Achieved 99.9% uptime through Grafana monitoring, performance tuning, and reliability engineering.
- Delivered a responsive React dashboard, increasing engagement by 40% and mobile performance by 20%.
Jun 2020 — Jan 2022 · Powai, Mumbai
LTIMindtree
Cloud Infra Engineer · Full-time
- Built automation and monitoring scripts, optimized SQL queries and database performance, and supported containerized deployments.
- Supported enterprise cloud infrastructure, incident troubleshooting, monitoring, and deployment workflows.
05 / Behind the code
A builder’s mindset.
A systems perspective.
I work where full-stack engineering, AI, and
distributed systems meet.
My focus is not just getting a feature to work. It’s making the whole system useful, understandable, and reliable— from data models and APIs to the interaction a user sees.
I’ve built for enterprise support workflows and US startups, bringing LLMs, retrieval pipelines, real-time communication, and automation into production.
My focus is not just getting a feature to work. It’s making the whole system useful, understandable, and reliable— from data models and APIs to the interaction a user sees.
I’ve built for enterprise support workflows and US startups, bringing LLMs, retrieval pipelines, real-time communication, and automation into production.
Own the outcome
Measure the improvement
Build for production
Education
Bachelor of Technology
Computer Science & Engineering
Netaji Subhash Engineering College
Kolkata, West Bengal · July 2016 — July 2020
Certifications
- Oracle Java Foundations Certification
- AWS Technical Professional
- IIT Madras — Python Data Structures and Algorithm
Based in
Bangalore, India
Experience across remote US teams and Indian technology companies.
06 / The next good problem
Great ideas deserve
great engineering.
Building an AI product, scaling a platform, or looking for an engineer who owns the outcome? Let’s start a conversation.
shubhankars361@gmail.com