Tanush.

Engineering High-Performance Systems & ML Pipelines.

Computer Science student at SLIET. Founder & CEO of Aetherin. Specializing in systems programming, memory quantization, vector search, and custom React architectures.

Get in Touch
Tanush Singla
0.28ms

Vector Query Latency

Custom SQ8 & ADC search

75%

Memory Reduction

8-bit quantization pipeline

LeetCode

Problem Author

"Largest Digit Product"

Founder & CEO

Aetherin SaaS

Leading tech & MVP build

Core Competencies

About & Proficiencies

Engineering robust systems from low-level memory management to high-throughput cloud applications.

Coding environment

Education

B.E. in Computer Science

Sant Longowal Institute of Engineering & Tech

Expected 2029

Diploma in Computer Science

Sant Longowal Institute of Engineering & Tech

2023 - 2026SGPA: 8.64

Technical Arsenal

C++

Expert

Impact • Low-level system architecture

Python

Advanced

Impact • Machine learning pipelines

JavaScript

Advanced

Impact • Frontend & Backend

TypeScript

Intermediate

Impact • Type-safe interfaces

SQL

Intermediate

Impact • Database querying

HTML/CSS

Advanced

Impact • UI development

Projects

A few things I’ve built — from low-level systems work to full-stack web apps.

Quantized Vector Search Engine

An 8-bit Scalar Quantization pipeline in Python that compresses float32 embeddings while maintaining semantic fidelity. Uses Asymmetric Dot Product search directly on compressed matrices with zero decompression overhead.

75% memory reduction · 0.28ms query latency · 100% Recall@10

PythonNumPyVector SearchQuantization

ML Stock Prediction Framework

End-to-end training and backtesting pipeline ingesting live market feeds. Benchmarks Linear Regression, Random Forest, and LSTM architectures with custom feature engineering and optimized memory allocations.

<15ms inference · ~40% faster pipeline after OOP refactor

PythonScikit-LearnLSTMPandasFlask

Madhuram Fest Website

Official web app for SLIET's cultural festival. Built with performance-first architecture — lazy-loaded assets, compressed images, and modular components designed for reuse across future editions.

1,000+ concurrent users at peak · ~30% smaller page assets

ReactNext.jsTailwind CSSVercel

Assistive Blind Stick

Arduino-based device for visually impaired users, using ultrasonic sensor arrays for real-time obstacle detection. Custom interrupt routines deliver immediate haptic feedback with minimal false positives.

Sub-100ms detection-to-feedback · field-tested with 5+ volunteers

ArduinoC++SensorsEmbedded Systems

Aetherin

A B2B SaaS platform I'm building for the education space.

As the founder, I lead a small development team across the entire product lifecycle — from architecture decisions and sprint planning to deployment pipelines. We're building tools that integrate AI capabilities into educational workflows.

The tech stack spans decoupled APIs, sentence embedding models, and LLM-powered features for content generation. I also manage a growing community of 500+ developers and educators around the product.

Aetherin - Brand Identity
Aetherin - Client Work
Aetherin - AI Content
Aetherin - Strategy
Get in Touch

Let's Connect

Have a project in mind, looking to hire, or want to collaborate? Reach out directly.

Send a Message

Contact Information

Looking to fill a software engineering or ML role? Feel free to copy my email directly or drop a message. I usually respond within a few business hours.