Ashwin Shenai

Ashwin Shenai

M. Sc. Computer Science

ETH Zürich

About Me

I am currently a Masters' student in Computer Science at ETH Zürich. Previously, I was a Data & Applied Scientist working with the INTL Ads Revenue & Relevance team at Microsoft India. I graduated from IIT Kanpur (IITK) with a Bachelor of Technology in Electrical Engineering in 2022. During my undergraduate years, I was an active member of Aerial Robotics IITK, where I managed and developed the software for our custom-built fleet of quadcopters.

Interests

  • Optimization & ML
  • Reinforcement Learning
  • Autonomous Systems
  • Algorithms

Education

  • B.Tech. in Electrical Engineering, 2022

    Indian Institute of Technology, Kanpur

  • M. Sc. Computer Science, Exp. 2025

    ETH Zürich

Publications

Proximal Algorithms for Smoothed Online Convex Optimization with Predictions

We propose receding horizon proximal-descent algorithms for solving smoothed online convex optimizations problems, which outperform the existing gradient-based algorithms.

Work Experience

 
 
 
 
 

Data & Applied Scientist

Microsoft India

Jun 2022 – Aug 2023 Bangalore, India
Designing, implementing, analyzing and tuning complex ML models residing at the core of the Microsoft paid search ecosystem. Key focus on leveraging actionable insights mined from large-scale data & signals from user query interactions to improve dense information retrieval techniques used for text ads selection.
 
 
 
 
 

Data & Applied Scientist Intern

Microsoft India

Jul 2021 – Aug 2021 Remote
Took a deep dive into dense information retrieval over the course of 6 weeks. Built state-of-the-art transformer-based retrieval systems and comparatively analysed their performance.

Projects

Everything I’ve worked on starting from my undergrad at IITK.

*

RL-based Internet Congestion Control

Course Project | EE698V: Advanced Topics in ML for Communication Networks, Fall 2022 | Experimented with RL-based algorithms for controlling the transmission of Internet packets

Molecular Communication via Diffusion with Multiple Transmitters and Receivers

Course Project | EE698U: Molecular Communication, Fall 2022 | Wrote a term paper discussing diffusion-based molecular communication systems in 3D

StackExchange Miner

Course Project | CS685: Data Mining, Fall 2021 | Developed a generalized framework to analyse, derive and present insights from the data dumps of every StackExchange site

Citation Network Analysis

Course Project | SOC471: Computational Sociology, Fall 2021 | Worked on gaining insights from the ArXiv citation dataset

Gesture Recognition

Course Project | CS698F: Sensing, Communication and Networking for Wireless Smart Devices, Fall 2021 | Developed a system to recognize and classify handmade gestures using a smartphone accelerometer and implemented many variants of the DTW algorithm

Swadeshi Microprocessor Challenge

Qualified for the Semi-Final Stage (National Top 100)

BOSCH’s Traffic Sign Recognition Challenge

Won the Gold Medal at this event in the 9th Inter IIT Tech Meet, IIT Guwahati

DRDO DGRE’s Vision-Based Obstacle Avoidance Drone Challenge

Won the Bronze Medal at this event in the 9th Inter IIT Tech Meet, IIT Guwahati

Graph-Based Modeling, Scheduling, and Verification for Intersection Management of Intelligent Vehicles

Course Project | CS637: Embedded and Cyber-Physical Systems, Fall 2020 | Implemented the aforementioned paper from EMSOFT 2019

Flipkart GRiD 2.0: Robotics Challenge - Indoor Autonomous Drone

Qualified Level-1 with Rank 1 in IITK, National Top 50 after Level 2

SDF Planning Library

Explored autonomous navigation and path planning based on TSDFs generated using Voxblox

Autonomous Navigation in Rough Terrain Environments

Mentored first-year students in implementing a navigation stack in simulation as a Summer Project

DRDO SASE’s UAV Fleet Challenge

Won the Gold Medal at this event in the 8th Inter IIT Tech Meet, IIT Roorkee

Beam Splitters and Single Photon Sources

Course Project | EE683: Quantum and Wave Phenomena, Fall 2019 | Modeled single photon sources in Python various emission distributions to verify the Hong-Ou-Mandel effect in simulation

International Micro Aerial Vehicle Challenge, 2019

One of the top 15 student teams internationally to participate in the Outdoor Problem Statement

Fastest Line Follower

Designed and built a line follower robot. Learnt the basics of manufacturing, electronics and controls

Pong: An AI Based Approach

Course Project | ESC101: Fundamentals of Computing [Advanced Track] Spring 2019 | Developed a Pong application with 1-player, 2-player interfaces and some strategic/RL agents