Lakshya Agarwal

Engineer by curiosity, philomath by nature

👋🏼 Welcome to my little corner on the Internet!

I'm currently a Software Engineer at Bain & Company. Previously, I built LLM-powered solutions as an AI Engineer in the education sector, and before that, I developed data-driven strategies for private equity clients at Bain.

Outside of work, you'll find me chasing high scores in Balatro or experimenting in the kitchen as I learn to cook - because, why not?

If you have an interesting project that you want to build with me, or ideas to explore together, I'd love to chat!

Selected Projects

View all projects
Pokemon Battling

A Pokemon battling game, implemented fully in TypeScript with an in-browser game engine and websocket support for real-time multiplayer

Sparse Autoencoder Explorer

A visual exploration of LLM feature activations by generating images on feature-adjusted outputs.

The Heist of Drake: Chapter 1

A choose-your-own-adventure game built using LangGraph, Flux, and Next.js in 24 hours

Micrograd

Tiny autograd engine with an interactive Streamlit demo to visualize backpropagation and a 3-class 2D classifier.

Video YouNiversity

AI-powered YouTube-based learning platform

WhatsApp Accident Bot

Automated WhatsApp bot for accident reporting and emergency response for endangered animals in Costa Rica

JD Interpreter

AI-powered job description analysis and resume optimization tool

LaunchLens

RAG-powered idea evaluation tool for Venture Capitalists (VCs)

Work Experience

Blog

Building RL environments for open-source AGI

Weekend notes on porting environments, running evaluations, and serving LMs on a cloud-hosted GPU.

Building a CLI Agent

My experience building a terminal-based AI agent and demystifying the hype around agents.

Vibe-coding a digital wardrobe

My thoughts on vibe-coding as of May 2025.

New Builds 2024

A recap of building a generative choose-your-own-adventure game in 24 hours at New Builds 2024 hackathon

Understanding the diffusion process

An exploration of the DDPM paper by Ho et al., covering core concepts of diffusion models for image generation.

LLM Evals

A tutorial on evaluating LLM performance using LangSmith for 10-K filing retrieval tasks.

LangChain Primer

An introduction to LangChain framework for AI applications, including RAG implementation and monitoring.

Fine-tune GPT-2 for Classification

A tutorial on fine-tuning GPT-2 for classification tasks using the Hugging Face Transformers library.

Building LSTMs from Scratch

A step-by-step implementation guide for building multi-layer LSTMs from scratch in Python.

GPT-4 Vision Example

A guide to using GPT-4 Vision API for generating advertising copy from product images.

Building a Transformer

A detailed breakdown of the Transformer architecture, focusing on encoder-decoder blocks and self-attention.

IMDb Rating Prediction

A regression analysis project predicting IMDb movie ratings and identifying key influential features.

EPL Season 14 Group D

An analysis of ESL Pro League Season 14 Group D performance metrics and player statistics.

Connect

Feel free to contact me at lakshyaag1202@gmail.com