Raunak Shukla
Building intelligent and scalable software systems.
About Me
Software & AI Engineer
Hey! I'm Raunak, a Software & AI Engineer with an MS in Computer Science from NYU. I build products end‑to‑end and ship fast. I'm currently working at Building Diagnostics Robotics, where I've designed a production ML pipeline for automated curve‑fitting and model evaluation and developed their with modular Next.js components and mobile‑first UX.
My work spans full‑stack (React/Next.js, Node.js, Django), mobile (React Native), and cloud (AWS/GCP, CI/CD) as well as applied ML/GenAI (TensorFlow, PyTorch, LLMs, fine‑tuning). I’ve built things like IntelliQuest (paper search engine), an AI Resume Builder using OpenAI, and big‑data pipelines on Spark and Snowflake.
Previously, I’ve delivered real‑time chat interfaces at Radical AI and supported 500+ learners at NYU’s Career Hub bootcamps. I care about clean, maintainable code and building scalable systems that make a real impact.
My Journey
From a curious CS student to a passionate software engineer - hover over the milestones to explore my journey.
Started CS Undergrad
Software Dev at YTBC
ML Intern at Feynn Labs
Started Masters at NYU
CS Grader at NYU
Bootcamp Admin
Started at BDR
Graduated NYU
SWE at BDR
Education
Master of Science (MS) in Computer Science
New York University - Tandon School of Engineering
Relevant Coursework
Bachelor of Technology (B.Tech) - Computer Science and Engineering
Vellore Institute of Technology
Relevant Coursework
Work Experience
AI Engineer
Building Diagnostics Robotics
•Engineered a retrieval-augmented generation (RAG) pipeline using OpenAI embeddings and Pinecone vector database to index and retrieve inspection and thermal analysis data, enabling context-aware inputs for downstream analysis.
•Deployed a dockerized ML pipeline on AWS Fargate to process large-scale thermal and drone telemetry data for anomaly detection and cost modeling, reducing runtime by ~40% through parallelization and resource optimization.
•Designed and implemented an LLM-based report generation pipeline to convert structured inspection and telemetry data into standardized natural language reports, enabling automated and consistent output generation.
•Optimized LLM performance through prompting strategies (few-shot, schema-constrained outputs) and implemented output validation and structured parsing to improve response consistency, reliability, and downstream system integration.
•Architected a cross-platform (Android & iOS) mobile application to capture site data and generate AI-powered inspection reports using the LLM pipeline with real-time data synchronization via GraphQL subscriptions, scalable to 1,000+ users.
Software Engineer Intern
Building Diagnostics Robotics
•Built and launched the company’s website from scratch, implementing server-side rendering, dynamic routing, and optimized content delivery, improving performance and page load times by ~30%.
•Designed a relational database schema to store user accounts and inspection reports, enabling secure, account-specific access through authenticated workflows, and integrated a payment system to support end-to-end transaction flows.
Bootcamp Admin
NYU Tandon Career Hub
•Oversaw LeetCode and Data Science bootcamps with ~500 participants across two semesters
•Managed logistics, automating data collection and analysis with Google Sheets
•Increased program retention by 20% through improved operational efficiency
Software Engineer Intern
Radical AI
•Built a real-time chat interface for the company's website with OpenAI API and scalable backend solutions using Node.js and Firebase-CLI
•Designed an activity dashboard for tracking user engagement using D3.js
•Conducted API testing and validation using Postman and optimized user experience with quick action buttons and default prompts
Grader for Introduction to Computer Science Course
NYU Courant
•Evaluated Java assignments and provided 1:1 feedback to students on their code to ensure best object-oriented practices
•Assisted students with debugging techniques and code optimization strategies
•Collaborated with professors to maintain consistent grading standards across multiple sections
Machine Learning Intern
Feynn Labs
•Modeled and fine-tuned machine learning models for market segmentation, achieving 92.2% accuracy by applying K-means clustering and hierarchical clustering on large-scale datasets
•Applied statistical modeling and data visualization on Tableau to support sales analytics and recommendations, enabling a startup to identify target customer segments and optimize marketing strategies
Software Developer
Young Tycoons Business Challenge
•Supervised a team of 5 members in building reusable components and key web pages for a production-level website using React.js
•Designed and developed a backend server with RESTful APIs to facilitate seamless integration with frontend
•Handled user authentication, data manipulation, and state management for over 20,000+ users
Projects
A curated set of projects spanning software engineering, AI, machine learning, and data engineering.
Multi-Agent Marketplace (Facebook Marketplace-style)
Engineered a multi-agent marketplace where autonomous buyer/seller agents search listings, exchange messages, and negotiate prices using LLM-driven decision making, powered by semantic search with OpenAI embeddings for context-aware retrieval.
Key Features
- •LLM-driven buyer/seller agents for autonomous search, messaging, and price negotiation
- •Real-time copilot chat with WebSockets and persistent session memory for end-to-end negotiation workflows
TaskPilot: Agentic Workflow Engine
Streamlined a multi-agent orchestration system enabling autonomous agents to plan, execute, and validate tasks with structured reasoning, contextual memory, and task coordination.
SOP Violation Slack Bot
Built a real-time retrieval-augmented generation (RAG) pipeline to monitor Slack conversations and detect policy violations by retrieving relevant SOP documents and evaluating compliance using LLM reasoning.
Reinforcement Learning for Complex Game Environments
Trained RL agents for CartPolev2, CarRacingv2, 3D Humanoid etc. using PPO, A2C, & DQN achieving multi-level task completion through stable and scalable training pipelines with Stable Baselines.
Analyzing Historical Temperature Anomalies
Conducted time-series analysis on the GISTEMP v4 dataset, implemented STL ARIMA and SARIMAX models, and developed a tuned RNN to project future temperature anomalies with a 6% error and an AIC of -183.
AI Resume Builder
A web app that parses PDF resumes and uses OpenAI API to generate or rewrite bullet points based on user-provided prompts or keywords.
Crypto Tracker
Real-time cryptocurrency tracking app with price alerts, portfolio management, and interactive charts for 10,000+ digital assets.
NYC Crime Data Analysis
Big data pipeline using Spark to analyze 218k+ NYC crime records with GCP, Airbyte, and Snowflake visualization.
IntelliQuest Search Engine
Academic paper search engine with advanced search options, related papers, and comprehensive database integration.
Research Publications
Published research work in machine learning, neural networks, and data analysis with contributions to academic conferences and journals.
Comparing Machine Learning Classification Algorithms and Feed Forward Neural Network to Perform Sentiment Analysis
Springer Conference - 6th ICSCSP 2023
Achieved an accuracy of 80% using a voting classifier and a feedforward neural network leveraging PyTorch to perform sentiment analysis on a dataset of 140k values.
Generating Stock Market Data and Making Predictions Using GAN and Neural Networks
Preprinted at Research Square
Generated artificial data using the TimeGAN model for 3 years and used this data for training neural networks. Trained RNN, LSTM & GRU neural networks & made predictions with MAE of 1.64 & MAPE as low as 1.53%
Total Revenue Prediction of A Sports Management Application: Grook Using Machine Learning Models
IEEE Conference - 13th ICCCNT 2022
Built an application for online booking of sports facilities using Java and used ML to predict revenue generated. Assembled a voting regressor and got best R² score of 0.979 and RMSE of 82 using polynomial features of degree 2.
Technical Skills
A comprehensive toolkit of technologies and frameworks I use to build scalable, efficient, and user-friendly applications.