// systems engineer — new delhi, india
Ankur Kumar Singh
Software Engineering · Distributed Systems
Backend engineer building low-latency, horizontally scalable systems. Distributed messaging, streaming pipelines, and the algorithms that make them fast.
- users onboarded in month one
- 5000+
- core-msg availability under failure
- 100%
- site load speed improvement
- 50%
- DSA problems solved · Knight
- 600+
about
I build low-latency, horizontally-scalable systems — distributed messaging, streaming pipelines, and the algorithms that keep them fast.
I'm a Software Engineering undergrad at DTU (B.Tech SE, 2027) focused on backend and systems work — Redis Pub/Sub, Kafka decoupling, WebSockets presence, fault-tolerant monorepos. I've shipped these in production as an AI Associate Intern at NMG Technologies and a web development intern at IIT Ropar (VLED Labs).
On the algorithmic side I hold Knight status on LeetCode (peak rating 1912, top 5%) and have solved 600+ problems across platforms — the same rigor I bring to distributed systems design and low-latency engineering.
I care about uptime, throughput, and clean failure modes: building systems that stay alive when infrastructure doesn't.
// experience
Where I've shipped
- 01 / 02
AI Associate Intern
NMG Technologies · New Delhi, India
Serverless platform engineering on Astro + Supabase for a centralized hackathon registration system.
Architected a unified registration platform replacing 5+ isolated event forms, onboarding 5000+ users in month one on infrastructure built for 50,000+ monthly registrations.
Automated 100% of user notifications across 3 lifecycle stages via Supabase PostgreSQL triggers, Edge Functions, and the Brevo API — saving 21+ hours of manual ops per event.
Accelerated site load speed by 50% and improved UX metrics by 30% with lightweight metadata placeholder frames served before asset hydration.
- 02 / 02
Web Development Intern
IIT Ropar (VLED Labs) · Remote
NPTEL internship on a core MERN-stack application — React, TypeScript, Express, Node.js.
Shipped 10+ responsive frontend features adopted across 6+ screens, translated into reusable architectures.
Integrated 12+ REST API endpoints with frontend clients for reliable state management and data hydration.
// projects
Things I've built
Selected systems — shipped, deployed, and running in production
- 01 / 06
Vortex: Distributed Real-Time Chat System
★ featuredHorizontally scalable real-time messaging on a Kafka + Redis backbone.
Distributed chat platform designed for scale: Redis Pub/Sub for cross-server Socket.IO broadcast, Apache Kafka decoupling high-throughput ingestion from PostgreSQL persistence, TurboRepo monorepo deployed across Vercel and Render.
Node.js · TypeScript · Socket.IO · Redis · Kafka · Prisma · PostgreSQL · Next.js
Sustained 100% core-messaging availability during Redis/Kafka outages via graceful fallback to direct broadcast.
Founded fault tolerance with graceful degradation across a TurboRepo monorepo.
- 02 / 06
Synapse: Research & Literature Discovery Platform
★ featuredFull-stack research platform with role-based access and a PubMed search pipeline.
Research collaboration platform with RBAC across 3 roles and JWT auth. Literature search integrates the NCBI PubMed API with a custom XML parser and caching layer.
React · TypeScript · Node.js · Express · MongoDB
Improved team collaboration efficiency by 40% with role-based access control (RBAC) and JWT auth.
Cut redundant upstream API calls by 60% via a custom PubMed XML parser and caching layer.
Boosted search recall by 35% on plain-language queries with a MeSH/UMLS query-expansion engine.
- 03 / 06
LoveAI: Persona-Adaptive AI Chatbot
Fine-tuned LLM that steers responses across 16 psychological personas.
Full-stack AI chatbot that derives user psychological personas from a 30-question matrix. Fine-tuned a Mistral base model with QLoRA (4-bit) + Hugging Face PEFT on 150,000+ dialogue pairs, with DPO emotional alignment and pgvector RAG.
React · Node.js · PostgreSQL · Mistral · QLoRA · AWS
Fine-tuned Mistral with QLoRA (4-bit) and PEFT on a 150,000+ pair corpus with DPO emotional alignment.
Integrated pgvector-backed RAG for dynamic context injection.
- 04 / 06
Omnichannel Customer Support Suite
Real-time multi-tenant helpdesk with an autonomous AI triage agent.
Enterprise helpdesk using WebSockets for zero-latency live messaging and presence. Multi-tenant Prisma schema enforcing RBAC and strict data isolation; LangChain agent triages tickets with semantic vector search for automated L1 resolution.
React · Node.js · PostgreSQL · WebSockets · Prisma · LangChain
Built real-time messaging + presence with WebSockets across concurrent sessions.
Enforced RBAC and tenant-data isolation with a multi-tenant Prisma schema.
Integrated a LangChain agent for automated Level-1 ticket triage via vector search.
- 05 / 06
Hermes-OpenClaw Agent Orchestrator
Agentic orchestration layer with autonomous tool-calling, deployed on DigitalOcean.
Engineered an agentic orchestration layer using local LLMs (Hermes/OpenClaw) with autonomous tool-calling and multi-step reasoning pipelines to resolve compound queries. Inference engine + agentic backend deployed on a DigitalOcean droplet with robust REST endpoints.
Python · DigitalOcean · LLM Agents · API Design
Implemented autonomous tool-calling and multi-step reasoning pipelines for compound queries.
Deployed inference on a DigitalOcean droplet exposing REST APIs for external clients.
GTM-focused modular architecture for fast adaptation across use cases.
- 06 / 06
Pocket FM — Strategic Analysis
In-depth market & retention case study for the audio OTT platform.
Comprehensive strategic case analysis for audio OTT platform Pocket FM: market penetration opportunities, user-retention strategies, and content-consumption trend evaluation for targeted growth.
Strategy · Market Research · Data Analysis
Evaluated market penetration opportunities and user-retention strategies.
Proposed actionable growth initiatives from content-consumption trend analysis.
skills
Languages
C++ · Python · TypeScript · JavaScript · SQL
Backend & Infra
Node.js · Express · Socket.IO · Kafka · Redis · Docker · AWS
Data & Storage
PostgreSQL · pgvector · MongoDB · Prisma · Supabase
Systems & CS
OS · Networks · DBMS · System Design · DSA · Low-Latency Systems
ML / AI
LangChain · Hugging Face · QLoRA · RAG
contact
Open to backend & systems engineering roles in quant / low-latency trading.