Chittapu Abhiram Reddy

Education

Chaitanya Bharathi Institute of Technology (Ongoing) Nov. 2022 – June 2026

B.E. Computer Science Engineering CGPA: 9.35 Hyderabad, India

Sri Chaitanya College June 2020 – May 2022

Intermediate Public Examinations: 97.5% | TSEamcet Rank: 2660 Hyderabad, India

SDE Intern - Providence June 2025 – Present

Software Development Engineer Intern Hyderabad, India

Designed an AI Agent for huge PDFs (containing >2500 pages, 20,000+ row tables), facilitating effective information retrieval from complex documents.
Implemented a custom preprocessing pipeline in snowflake to hide Personally Identifiable Information(PII) and Protected Health Information(PHI) to achieve HIPAA compliance.
Designed a RAG-based chatbot to deliver accurate, context-aware responses on organizational tools, processes and resource access to improve employee onboarding experience.

MEDRAGA - Medical Assistant LLM | Python, Langchain, FastAPI, Cohere, Playwright

Link

Developed a medical assistant that performs up-to 35% better than the competition in providing personalized treatment plans and second opinions.
Utilized Retrieval Augmented Generation (RAG) to feed the model with cutting-edge medical research & patient data for better-informed predictions.
Leveraged Qdrant vector database for fast retrieval of relevant data and Langchain to create the RAG pipeline.

MisclassifyMe - Fooling Image Classifiers | Pytorch, TorchVision, Pillow

Link

Engineered FGSM (Fast Gradient Sign Method) and PGD (Projected Gradient Descent) based algorithm to generate adversarial images to fool image classifiers to misclassify target images.
Generated Adversarial images with imperceptible changes using PGD which were misclassified by the VGG16 model with up to 95% confidence.
Simulated targeted attacks using PGD causing the VGG16 classifier to misclassify images to a chosen target class with over 85% reliability.

Verisite - Verifying website authenticity | Python, Pytorch, XGBoost, FastAPI, Transformers, QLora

Link

Built an ML-based browser extension to detect malicious/phishing websites with 83% accuracy.
Fine-tuned a DeBERTa-v3 language model using qlora to distinguish between benign and malicious websites.
Added a multi-input neural network on top of the DeBERTa model to improve precision by 3%, greatly decreasing the false positive rate.

Languages: Python, C, C++

Web & Backend: React, FastAPI, GitHub

Databases: MySQL, MongoDB

ML Libraries & Frameworks: TensorFlow, PyTorch, Keras, Scikit-learn, LangChain, Langgraph

Competitive Programming

Machine Learning Specialization

by Stanford and DeepLearning.ai

Completed the Machine Learning Specialization by Stanford university on Coursera covering the theoretical and practical aspects of machine learning

Vice President, HackItOn, CBIT February 2024 – Present

Spearheaded the organization of multiple large-scale hackathons with a price pool of up to rupees 1.2 Crore, attracting over 500 teams from various institutions.