Hello, I'm
MS CS (AI) @ SUNY Binghamton · Building AI products at Bytecrafts · OSS: langgraph-ui-components
I build web apps, ML systems, and developer tools — from React/Next frontends to Rust & Python backends and AI-powered pipelines.
0+
Projects Shipped
0+
npm downloads/week
0+
Years Experience
0+
Open Source Repos
About Me
Hi, I'm Karthik — MS CS (AI) student at SUNY Binghamton. I build production full-stack systems and AI-powered products: currently developing a 3D model generation SaaS at Bytecrafts using React + LangGraph, and running Career Pipeline — a self-hosted job aggregation platform on Docker + Kubernetes. I authored langgraph-ui-components, an open-source React library for AI chat UIs averaging 220+ npm downloads/week. My approach: visualize the system first, talk to real users second, then build.
My Projects
Career Assistant — Side Panel Extension (MV3)
A Manifest V3 browser extension with a Side Panel UI for sign-in (Email/Google via Firebase), LaTeX resume ZIP upload, per-user generation history in Firestore, and one-click Resume + Cover Letter PDF downloads. Built with Vite + React + TypeScript, Tailwind CSS, and Firebase Auth/Firestore/Storage.
Social Media Analytics Platform
Cross-platform analytics comparing engagement on 4chan and Reddit with AI-powered insights (LangChain + CopilotKit). FastAPI + PostgreSQL/TimescaleDB backend; Next.js + TypeScript frontend with ApexCharts dashboards. Includes toxicity scoring (Google Perspective API), Faktory background workers, and a natural-language SQL agent.
Resume Tailoring Agent
LangGraph-orchestrated AI agent that auto-tailors resumes to job descriptions and generates ATS-optimized 1-page PDFs. Multi-step pipeline: JD analysis, company web research, batched project scoring, LaTeX section generation. Supports 4+ LLM providers (OpenAI, Gemini, Moonshot) with graceful degradation. 99.8% pdflatex compilation success rate.
LangGraph UI Components
Open-source React component library for building AI chat interfaces with LangChain/LangGraph integration. Drop-in UI primitives — graph nodes, edges, state panels — that plug directly into LangGraph-powered applications. Published on npm with 220+ avg weekly downloads.
Reddit & 4chan Knowledge-Graph RAG Index
Retrieval-Augmented Generation system that builds a Neo4j knowledge graph from Reddit and 4chan data and enables graph-aware QA using LlamaIndex, LangChain, and CopilotKit. Backend uses FastAPI + PostgreSQL; graph storage in Neo4j; LLMs via OpenAI.
AI Job Analyzer — Resume Screening & Live Job Matching
Cloud-native AI resume screening and live job matching using Sentence-BERT for semantic similarity, Streamlit UI, LinkedIn scraping (Playwright + BeautifulSoup), and AWS deployable via Terraform with ALB and autoscaling.
Placement Prediction
Web-based placement prediction system using Django and machine learning. Classification model predicts student placement outcomes based on CGPA, internships, and certifications. Algorithms include Random Forest and CatBoost, evaluated using accuracy, precision, and F1-score. Achieved 86% accuracy.
Fake News Classification Using LSTM
LSTM model for detecting and classifying news articles as real or fake. Dataset of 44,000+ articles from Reuters and unreliable sources. Implements NLP preprocessing with SpaCy and NLTK. Achieved 85% accuracy in fake news detection.
Hashassin — Password Generator & Hash Cracking (Rust)
Rust CLI for generating random passwords, creating cryptographic hashes, building rainbow tables, and cracking hashes. Includes a server-client mode for uploading hash files and tables. Supports MD5, SHA-family, scrypt, and parallel cracking via rayon and tokio.
Flask Chat Application (WebSockets + MongoDB)
Real-time Flask chat application using WebSockets with MongoDB persistence. Dockerized and deployed on Kubernetes with 3-replica Flask deployment for high availability. Supports real-time messaging and message storage.
Kabaddi Event Management App
Full-stack Kabaddi League management platform. Users can book events and view/edit bookings; admins control referees, teams, and venues. Built with React + Spring Boot, Spring Security JWT auth, MySQL, and Axios.
Drug Sensitivity Prediction Using Multi-Modal Data
Predicts drug response (LN_IC50) for cancer cell lines using multi-modal GDSC data. Pipeline includes imputation (KNN, RF), encoding strategies, XGBoost regression with RandomizedSearchCV, and SHAP-based interpretability.
Lane and Vehicle Detection
Computer vision system for lane detection using OpenCV. Isolates the Region of Interest in road images using polygon masking and edge detection. Essential technique for self-driving car navigation pipelines.
NYC Motor Vehicle Collisions Data Analysis
Data analysis of NYC Motor Vehicle Collisions: cleaning, preprocessing, time-series and geospatial exploration. Identified Brooklyn (35% of crashes) and 16:00–17:00 as peak crash time. Visualized with Pandas, Matplotlib, and Seaborn.
Data Collection System — 4chan & Reddit
Continuously ingests posts from 4chan boards and Reddit subreddits into PostgreSQL with Faktory background workers. Features incremental crawling, toxicity scoring via Google Perspective API, and reproducible Python environment with ruff linting.
Mitra — Candidate Onboarding Platform
Onboarding platform for Swizy, Zomato, and Dunzo delivery partners. Candidates log in, set preferences, and track performance. Admins onboard candidates, assign vendors, and monitor performance. Built with ReactJS + Spring Boot.
Online Quiz
University website for online quizzes and assignment submissions. Students log in, take quizzes, and track progress. Built with PHP, HTML/CSS, Bootstrap for responsive design, and jQuery for interactive elements.
Towards Light
Platform for connecting with Reiki healers and psychics for personalized healing and guidance. Book sessions for clarity, stress release, and balance. Seamless booking experience built with modern web technologies.
Web Portfolio
Personal web portfolio showcasing projects, skills, and achievements. Built with Next.js 14 App Router, Tailwind CSS, Framer Motion animations, and GSAP. Features animated skill grids, project filtering, and a Calendly integration.
Adventures
Beyond Code
When I'm not pushing commits, I'm pushing elevation. Hiking trails taught me patience, planning, and that sometimes the best views require the hardest climbs — lessons that carry over perfectly to engineering.
Badminton
Fast-twitch thinking — on court & on screen
What Co-workers &
Friends Say
Swipe the cards or use the arrows to hear what people I've worked and hacked with have to say.
Lokesh Kumar Saspu
Senior SWE @ Virtusa Consulting Pvt Ltd
“Karthik has an impressive ability to ship fast without sacrificing quality. Worked alongside him on a RAG pipeline — his systems thinking is top-notch.”
Pruthvi Kumar
Founder @ Digiphins Pvt Ltd
“One of the most curious engineers I've met. Karthik doesn't just build features — he understands why they matter and optimizes for real user impact.”
Gaurav
Founder @ Bytecrafts
“Worked with Karthik on an open‑source React component library for LangGraph‑based AI chat interfaces. His work on graph nodes, edges, and state panels helped create a clean, reusable UI layer, and he was a professional, collaborative teammate.””
Let's Connect
I'm currently looking for new opportunities — my inbox is always open. Whether you have a question or just want to say hi, I'll get back to you!
Location
New York, USA