Hi, I'm Seshu Swaraj

AI Engineer specializing in Foundation Models, Generative AI, and Deep Learning. Currently building Large Wireless Models (LWM) at Ericsson, working with advanced architectures like Transformers, Mamba, and Hyena to solve real-world wireless communication problems.

I design and develop AI systems including LLM-based applications, predictive models, and data-driven solutions.

View Resume View Work

What I Do

🚀

AI Systems Builder

I build end-to-end AI systems — from data processing to model deployment — solving real-world problems using machine learning and deep learning.

🧠

Foundation Model Engineer

Working on Large Wireless Models (LWM) at Ericsson, experimenting with advanced architectures like Transformers, Mamba, and Hyena for wireless intelligence.

✨

Generative AI Developer

I create LLM-powered applications like chatbots and intelligent assistants using RAG pipelines, LangChain, and vector databases.

📊

Data-Driven Problem Solver

I develop predictive models for real-world use cases like churn prediction, stock forecasting, and time series analysis.

Experience

Data Science Intern – Foundation Models

Ericsson, India

Jun 2025 – Present

Contributing to the development and optimization of Large Wireless Models (LWM) for representation learning on wireless channel state information (CSI) data. Designed and implemented deep learning pipelines in PyTorch for tasks including LoS/NLoS classification, beam prediction, channel estimation, and channel interpolation. Experimented with multiple foundation model architectures such as Transformer baselines, Mamba, Hyena, Falcon, RadioLLM, and WCFM to evaluate performance across wireless learning tasks. Worked on model optimization and benchmarking to improve efficiency and prediction accuracy across different architectures. Generated and processed large-scale datasets using DeepMIMO and Sionna frameworks, enabling robust training and evaluation in realistic wireless communication environments.

Python PyTorch Foundation Models Transformers Mamba Hyena Falcon RadioLLM WCFM DeepMIMO Sionna Wireless CSI

Java Full Stack Intern

Kodnest Technologies, Bengaluru

Aug 2023 – Mar 2024

Trained and worked on Java full stack development with focus on backend fundamentals. Practiced Spring Boot concepts, REST API development, and MySQL database operations through mini applications and structured coding modules.

Java Spring Boot REST APIs MySQL

Education

M.E. in Artificial Intelligence & Machine Learning

Manipal School of Information Sciences, MAHE

Jun 2024 – May 2026

CGPA: 7.44

B.Tech in Computer Science & Engineering

GIET College of Engineering

2019 – 2023

CGPA: 7.23

Core Expertise

Generative AI & LLM Systems

Built real-world AI systems like chatbots & semantic search using LLM pipelines

RAG Pipelines (LangChain + LlamaIndex)

LLMs & Transformers

Vector Search (ChromaDB)

Machine Learning & Prediction Systems

Developed predictive models for churn, stock trends, and forecasting applications

Model Development (Scikit-learn, SVM)

Time Series (ARIMA, LSTM)

Feature Engineering & Optimization

Deep Learning & AI Models

Applied deep learning for medical image analysis and classification tasks

PyTorch & TensorFlow

CNN, RNN, LSTM Architectures

Programming & Data Handling

Strong foundation in coding, data processing, and backend logic

Python (Core + ML Stack)

SQL & Data Querying

Java (OOP & Application Dev)

Tools, Deployment & Visualization

Built interactive AI apps, dashboards, and deployable solutions

Streamlit (AI Apps)

Power BI (Dashboards)

Git & GitHub

Certifications

IBM Generative AI & LLM Engineering Specialization

IBM • MAHE Manipal
Covers Transformers, LLM Architecture, Fine-Tuning, NLP, Prompt Engineering & Data Preparation.

Java Full Stack Development

Kodnest Technologies • 9 Months

Programming in Python

Meta • MAHE Manipal

Machine Learning with Python

IBM • MAHE Manipal

Generative AI with Large Language Models

DeepLearning.AI • MAHE Manipal

LangChain & Prompt Engineering

DeepLearning.AI • Covers LangChain, Prompt Engineering (Llama 2 & 3)

Agentic AI & AI Agents

Vanderbilt University • MAHE Manipal

Time Series & Advanced ML Models

IBM • Covers Time Series Forecasting & Survival Analysis

SQL for Data Analysis

Coursera • Includes Aggregations & Querying Data

Featured Projects

🤖

BotPYT AI

LLM-powered chatbot enabling context-based Q&A from PDF documents and YouTube transcripts with semantic search using ChromaDB.

Python LlamaIndex ChromaDB Streamlit

✉️

Cold Email Generator

Personalized outreach email generation using LLM prompt-driven workflows with LangChain and Streamlit interface.

Python LangChain Llama 3.1 Streamlit

💰

Bank Churn Prediction

ML-based predictive model using SVM to identify customers at risk of leaving, with comprehensive data preprocessing.

Python Scikit-learn SVM ML

🏥

Hospital ER Dashboard

Real-time Power BI dashboard providing visual insights into emergency room performance with KPI analytics.

Power BI Data Viz KPI

🏧

Virtual Teller Machine

Java-based ATM simulation supporting core banking operations with interactive GUI and OOP principles.

Java OOP GUI

🏏

IPL Analysis Dashboard

Interactive Power BI dashboard analyzing IPL match performance, player stats, and team insights with real-time visual analytics.

Power BI Data Visualization

🛒

Blinkit Sales Dashboard

Built a real-time analytics dashboard to track grocery sales trends, customer behavior, and operational KPIs.

Power BI KPI Analytics

🧬

Cancer Detection using GNN

Implemented Graph Neural Networks to detect Squamous Cell Carcinoma from histopathological images with 82% accuracy, addressing class imbalance and improving model performance.

Python Deep Learning GNN

📊

Stock Trend Prediction

Time series forecasting model using ARIMA, LSTM, and moving averages to predict stock price trends with real-time data from yFinance.

Python Time Series LSTM

🌍

Climate Change Forecasting

Developed predictive models using ARIMA, SARIMA, LSTM, and Prophet to analyze temperature and CO₂ trends.

Python Time Series

🥇

Gold Price Forecasting

Forecasted gold prices using ARIMA, LSTM, and Prophet with advanced preprocessing and visualization techniques.

Python ML

📅

Doctor Appointment Web App

Responsive web application for booking and managing doctor appointments with dynamic scheduling and user-friendly UI.

HTML CSS JavaScript

Achievements

🥇

5th Rank (Team) in the ITU LWM Challenge - Global competition on Large Wireless Models

⭐

Awarded Most Valuable Player in Cyber Security Performance, 2019

📢

Selected to deliver a talk on Java Programming during undergraduate studies

🏅

Secured 5th Rank in undergraduate college examinations