AARADHYA
SAXENA
Applied AI engineer building systems that turn messy human intent into useful decisions - across search, recommendations, personalization, and LLM-powered products.
Particularly drawn to healthcare, human performance, and systems that improve how people live day to day.
SELECTED_WORK
01 / 04
JioMart & Tira Beauty :: Search & Recommendation
2022 — 2024Led the development of a real-time personalization and ranking system for JioMart, one of India’s largest grocery platforms. Designed systems to serve millions of user sessions with high-throughput retrieval and low-latency ranking.
Built the product recommendation system for Tira Beauty, powering product discovery across 4.3M monthly users through session-aware ranking and recommendation pipelines.
Developed a centralized search service used across multiple Reliance e-commerce platforms, enabling unified retrieval and ranking across domains.
02 / 04
Estetica :: Wayfair AI Sales Assistant
2024 — 2025Built an LLM-powered conversational sales assistant for Wayfair that replaces static product recommendations with dynamic, intent-driven product discovery.
Unlike traditional “More Like This" systems, Estetica enables users to iteratively refine queries (e.g. “same product in a slightly darker shade"), translating natural language into structured retrieval and ranking pipelines in real time.
Won 2nd place among 200+ teams at Wayfair India TDC hackathon, then productionized for launch on Wayfair's platform.
03 / 04
Bettermeals :: Jarvis for Kitchen
2025 — PresentBuilding an AI personal assistant that automates all things health. Designed as a “Jarvis for the kitchen", translating user health data, preferences, and constraints into actionable meal decisions — including doctor syncing for chronic conditions like diabetes.
Multi-agent pipeline combining LLM reasoning, constraint-based optimization, and real-world execution systems (cook workflows, shopping automation).
04 / 04
SOFIE :: C++ Inference Engine
2021Built and extended SOFIE, a code-generation based inference engine that converts ONNX models into standalone C++ for high-performance execution. Designed for low-latency inference in high-energy physics workloads at CERN.
Implemented core ML operators within the codegen framework and optimized execution paths using BLAS. Published at Journal of Physics 2021.
PROJECTS_
&_HACKS
Research, hackathon builds, and experimental systems across privacy, healthcare, and computer vision.
$ projects --domain privacy health vision games
> Nights_Watch [CHROME_EXTENSION]
> Cheenikum_AI [GDM_COMPANION]
> Optical_Flow_Estimation [RESEARCH]
> YuMe [GENAI_HACKATHON]
> The_Martian [ED_TECH_GAME]
$ hackathon --list --winner-only
> Google_GenAI_Hackathon [1ST_PLACE]
> South_Park_Commons [1ST_PLACE]
> Groq_AI_Hackathon [1ST_PLACE]
> Wayfair_TDC [2ND_PLACE]
> Wayfair_GST [TOP 5]
The Martian
A space survival game where players learn science, systems thinking, and coding through a ten-mission rescue campaign.
Night's Watch
Built a Chrome extension that protects users from accidentally pasting sensitive information into AI chat tools. Detects clipboard content, masks common PII patterns, and inserts a security layer between the clipboard and supported GenAI interfaces.
Cheenikum AI
Built a gestational-diabetes companion focused on blood glucose tracking, personalized meal guidance, and emotional support during pregnancy. Designed to translate health inputs into practical, culturally aligned daily care workflows.
Unsupervised Optical Flow Estimation
Research project under Prof. Bhabatosh Chanda on monocular optical flow estimation. Proposed an unsupervised loss formulation for learning dense motion without ground-truth labels, with downstream application to drone collision prevention.
YuMe
Built a personalized health and wellness companion for the Google Gen AI Exchange ecosystem. Positioned around guided nutrition and personalized user support, with a strong consumer-facing product angle.
Career_Log
Bettermeals
Founder
Built BetterCare, WhatsApp-native multi-agent care-continuity platform that lets clinics, pharma, and wellness brands run personalized patient/customer programs and contextual recommendations driven by tenant/program-specific templates.
Building a personal assistant for health and the kitchen, designed to make everyday eating more intelligent, personalized, and automatic.
Wayfair
Senior Software Engineer - Machine Learning
Spearheaded Estetica, Wayfair's AI sales agent, for conversational product discovery; also deployed in-storefront to surface alternatives and similar products in real time, cutting time-to-conversion and reducing drop-off to lift conversion by ~20%.
Built and maintained core pricing infrastructure for price updates and ran pricing experiments across 4 mil SKUs.
People+ai
Volunteer
Experimented with Cheenikum.ai. Intelligent companion app for Gestational Diabetes that helps expecting mothers track meals, monitor glucose, and get personalised, culturally-relevant health insights in real time. https://www.producthunt.com/posts/cheenikum-ai
Jio Platforms
Backend-ML Engineer / Software Engineer - ML
Led the development of the in-house Personalization system for JioMart search from scratch. India's 2nd-largest online grocery platform, serving ~600K orders/day across 260+ cities.
Built the product recommendation system for Tira Beauty. (Around 4.3 million monthly visits)
needl.ai
Software Engineer
Full-stack engineer focused on product usability, dashboard feature development, and third-party integrations.
CERN
Software Engineer · European Organisation for Nuclear Research
Built SOFIE, a code-generation inference engine that compiles ONNX models into dependency-free, BLAS-optimized C++ for low-latency inference at CERN; open-sourced in ROOT, published at ACAT 2021.
Education
Indian Institute of Technology Roorkee (IIT Roorkee)
BTech+MTech