Ranjan Singh

I'm a Gen AI & Product Engineer with nearly a decade of experience shipping production AI and user-facing systems across web and mobile.

I'm currently Lead Engineer at Telos in New York, where I took the AI Life Coach platform from early prototype to production across iOS, Android, and web, leading the engineering team across architecture, delivery, and product analytics. I architected the multi-agent LLM system (LangChain + LangGraph) behind the journaling and coaching experience, with long-term memory via Zep and retrieval grounded in pgvector.

The part I care most about is what happens to your data on the way to the model. At Telos I built a privacy-by-design pipeline using Microsoft Presidio to de-identify user content before any prompt leaves the platform, so zero PII reaches third-party LLMs. That problem turned out to be much bigger than one product, which is why I'm now building Omit: local-only PII redaction that runs entirely on your machine, with no cloud, no accounts, and no telemetry.

Day to day I work in TypeScript, Python, React, Next.js, React Native, and Node.js, and I care a lot about the things that don't show up in a demo: latency under load, failure modes, and code other people can maintain.

Currently working on

Omit: offline, privacy-first PII redaction for Windows.
AI powered Journal for Intentional Living: Telos.
Shelly: a floating terminal companion for AI coding workflows.
My personal site ranjansingh.dev.

Timeline

2024-2025

Architected and launched Telos, an AI-powered journal platform for intentional living (https://heytelos.com/).
Engineered a high-performance backend utilizing Server-Sent Events (SSE) and auto-scaling with Cloud Run, reducing response latency by 68% and supporting 1,000+ users with 99.9% uptime.
Designed and implemented an intelligent Chat Agent with Retrieval-Augmented Generation (RAG) capabilities using Langchain and multiple LLM providers including Claude and OpenAI, improving user engagement by 40% and reducing response time to under 2 seconds.
Contributed to Microsoft's Presidio open-source project, enhancing PII data de-identification capabilities with 2 merged pull requests improving accuracy by 5%.

2022-2023

Engineered an AI-powered personalized trip itinerary generator (Quick Start), serving 20,000+ travelers and reducing initial planning time by 70% through intelligent automation and user preference analysis.
Automated email attachment processing for travel planning workflows, eliminating 25+ hours of manual work per week.
Implemented AI-driven booking import functionality for transportation and accommodation, streamlining the reservation management process and reducing data entry errors by 80%.
Strengthened application reliability by implementing comprehensive error boundaries and Sentry monitoring, reducing crash rates by 45% and improving overall stability to 99.8% uptime.
Revamped backend architecture to boost concurrency, achieving a 3x scalability increase and slashing API response times by over 75%, from 8 seconds down to under 2 seconds.
Developed sophisticated internal tools for streamlined application management and comprehensive performance monitoring, reducing debugging time by 60% and improving deployment speed by 40%.
Completed advanced certifications in Software Architecture & Technology of Large-Scale Systems and Artificial Intelligence A-Z™, expanding technical expertise in enterprise-scale solutions.

2021

Architected and developed a robust Node.js backend using Express.js, successfully processing 20,000+ legal cases weekly across multiple law firms with 99.5% accuracy and zero data loss.
Enhanced system communication efficiency by implementing RabbitMQ message queuing between Crawler and Backend services, reducing pending request backlogs by 80% and improving processing speed by 3x.
Created and published sharp-multer NPM package, providing automated image resizing, optimization, and watermarking capabilities for multer file uploads, now used in production by multiple downstream projects.
Integrated real-time in-app notifications using Socket.io, improving user engagement by 35% and reducing notification delivery time to under 500ms for 5,000+ concurrent users.
Successfully migrated React Native application from legacy version 0.53 to 0.60, ensuring compatibility with modern devices and achieving 30% performance improvement while maintaining 100% feature parity.
Developed and published custom React Native package Country Code Picker, a lightweight country code selector filling a gap in the React Native component ecosystem.

2020

Developed a comprehensive e-commerce platform using Next.js and Material-UI for Revo Telecom Parts Inc, managing 50,000+ SKUs with advanced product catalog functionality and achieving 40% increase in conversion rates.
Significantly improved SEO performance by implementing multi-language support across the entire product catalog, enhancing global market reach and increasing organic traffic by 150% across 8 international markets.
Engineered intelligent auto-localization features supporting 8 languages without relying on IP-based detection, improving user accessibility and experience while reducing bounce rate by 25%.
Architected and developed both native mobile application and backend infrastructure for wedding marketplace Wedcell, creating a comprehensive platform serving 2,000+ wedding service providers and 10,000+ couples with 95% user satisfaction rate.
Mastered modern web technologies including ReactJS, TypeScript, Redux, and Ant Design, expanding full-stack development capabilities and reducing development time by 45%.

2018 - 2019

Implemented advanced GPS-based location tracking solutions using React Native and Firebase, delivering precise real-time positioning capabilities with 3-meter accuracy for 1,000+ active devices.
Developed offline-first location tracking architecture, ensuring reliable data collection even in low-connectivity environments and achieving 99% data capture rate across 500+ remote locations.
Integrated Hermes JavaScript engine to optimize React Native application startup performance, reducing initial boot time by 60% (from 8 seconds to 3 seconds) and improving user experience metrics.
Built a sophisticated native IoT control application with Bluetooth and WiFi connectivity, enabling seamless device management and monitoring for 200+ connected devices with 99.2% uptime.
Designed and developed a 2D endless runner mobile game featuring custom character design using Unity Game Engine, achieving 50,000+ downloads and 4.2/5 app store rating, demonstrating versatility in game development.