# Shikhar Agnihotri

Senior Software Engineer working across backend systems, product infrastructure, and applied AI.

Location: Bellevue, WA
Current: Salesforce
Education: Carnegie Mellon University, School of Computer Science

Links:
- Website: https://shikhar-scs.github.io/hey/
- Resume: https://shikhar-scs.github.io/hey/ShikharAgnihotri.pdf
- Email: mailto:shikhar9a@gmail.com
- GitHub: https://github.com/shikhar-scs
- LinkedIn: https://www.linkedin.com/in/shikhar9a/

## Introduction

I am a software engineer at Salesforce. I build backend systems and applied AI products.

I am drawn to work where better engineering makes a visible difference to the person using it.

## Selected Work

### InvestInsights.ai
- Context: 2025 / AI product
- Summary: An AI-assisted investment research product that brings filings, market events, and web context into one reviewable workflow.
- Why I built it: Investment research is fragmented across portfolios, filings, market events, and news. I built InvestInsights to bring that context into one workflow and make it conversational.
- The interesting problem: Adding a chat interface is straightforward. Choosing the right context, keeping responses grounded, and presenting uncertainty clearly is the real product and engineering work.
- Tags: LLMs, MCP, React Native
- Link: https://investinsights.ai/

### Google Meet Classic Impersonator
- Context: 2020 / Chrome extension
- Summary: A Chrome extension that replaced live Google Meet video with animated characters using client-side TensorFlow.js and SVG rendering.
- An experiment that escaped the prototype: During the shift to remote meetings, I wanted to replace live video with animated characters. The extension ran pose estimation in the browser and translated movement into SVG animations.
- What made it interesting: What began as a playful experiment became an exercise in fitting real-time machine learning into the constraints of a browser extension.
- Tags: Chrome Extension, TensorFlow.js, SVG
- Links: https://chromewebstore.google.com/detail/google-meet-classic-imper/eojkkpaoohebkdkplhjjcedcpkajlibe, https://www.youtube.com/watch?v=RvUI0IsnkeE

### Speech enhancement over networks
- Context: 2023 / Research
- Summary: CMU research on speech enhancement under network and background noise, evaluated through perceptual quality and intelligibility.
- Research under real-world constraints: Speech models are often evaluated on clean, controlled inputs. This work examined what happens when network degradation and background noise become part of the problem.
- What we measured: The evaluation considered both perceptual quality and intelligibility rather than treating a cleaner signal as the only measure of success.
- Tags: PyTorch, Signal processing
- Link: https://arxiv.org/pdf/2303.09048.pdf

## Additional Projects

- Search over messy information: coursework comparing classic, dense, and LLM-assisted retrieval across ranking and evaluation tradeoffs. Tags: Python, Lucene, Faiss, LLMs.
- Twitter ETL and Analytics: scalable cloud service for querying a large tweet corpus. Tags: AWS, Kubernetes, Java, Helm.
- Adversarial attacks on LLMs: course research on adversarial prompts and red-team data generation. Tags: PyTorch, Python, LLMs.
- Walket: routing and batching prototype for parallel warehouse fulfillment. Link: https://shikhar-scs.github.io/bayesian-retails/
- AoIP.ai: SDK work around peer-to-peer VoIP, AI data generation, and scalable communication infrastructure. Links: https://github.com/KonanAI/aoip.ai, https://drive.google.com/file/d/1NUPdqnY3f7Lc8ogv5nYA5F3v040mKHKt/view?usp=sharing
- Stuffer: Alexa skill built on AWS Lambda.
- YouTube Views Enhancer: early Chrome Tab APIs experiment. Link: https://github.com/shikhar-scs/Youtube-Views-Enhancer#youtube-views-enhancer-

## About

I like building software that people use to get real work done. That has meant order fulfillment at Walmart, technical interviews at HackerRank, and messaging infrastructure at Salesforce. AI is now part of my day-to-day workflow for understanding code, comparing approaches, and reviewing changes; I still own the design and final implementation.

## Experience

### Salesforce
- Role: Software Engineer, MTS to Senior Software Engineer, SMTS
- Tenure: 2023 - Present, internship + full-time
- Summary: Enterprise messaging infrastructure across APIs, diagnostic tooling, reliability, and full-stack product surfaces.

### D. E. Shaw
- Role: Software Engineer, Member of Technical Staff
- Tenure: 2022, full-time
- Summary: Backend workflow orchestration, with a focus on fault recovery and operational automation.

### Walmart
- Role: Software Engineer Intern to Software Engineer II
- Tenure: 2019 - 2022, internship + full-time
- Summary: Distributed fulfillment systems built around resilient APIs, cloud services, and production operations.

### HackerRank
- Role: Software Development Intern
- Tenure: 2018, internship
- Summary: Real-time collaboration features for developer interviews, including video and shared development environments.

## Education

### Carnegie Mellon University
M.S. in Intelligent Information Systems from the School of Computer Science, followed by a year teaching CMU's Intro to Deep Learning course as a graduate teaching assistant.

### NSIT Delhi
Bachelor's in Computer Science, with early work across NLP, language resources, and software engineering.

## Contact

Email: shikhar9a@gmail.com