Software Engineer · AI/ML Systems · Buffalo, NY

I build reliable software for intelligent systems.

Graduate Computer Science researcher at the University at Buffalo working across Python software, retrieval systems, computer vision, machine learning pipelines, and cloud-based applications.

View selected work Contact me

80%processing throughput improvement

95.3%worst-group accuracy

3.958 / 4graduate GPA

Experience

Engineering work with measurable outcomes.

I focus on the software around models as much as the models themselves: reusable components, data pipelines, evaluation, retrieval, performance, and maintainability.

Jan 2026 — Present Buffalo, NY

Research Assistant — AI Software Development

University at Buffalo, SUNY

Research Engineering

Developed a modular assistive-navigation pipeline in Python and PyTorch around a frozen DINOv2 vision encoder, semantic-segmentation decoder, and task-specific components.
Built reusable modules for ADE20K data loading, preprocessing, training, validation, checkpointing, and inference to support reproducible experimentation.
Improved training and evaluation workflows with mixed precision, automated checkpoint selection, and quantitative monitoring, reaching 50.1% validation mIoU and 83.2% pixel accuracy.
Designed extensible interfaces for future intent-routing and vision-language integration; the work contributed to a competitively selected abstract presented at the Stony Brook AI Symposium.

PythonPyTorchDINOv2Computer VisionML Pipelines

Jan 2025 — Jun 2025 Remote · India

Software Development Intern

Excl Solutions Pvt. Ltd.

Python · Retrieval

Developed and optimized end-to-end Python image-processing pipelines for large-scale TEM image analysis, improving processing throughput by up to 80%.
Built an embedding-based semantic retrieval application for natural-language search across internal datasets and technical documents.
Implemented modular components for ingestion, preprocessing, model inference, vector retrieval, and result generation to improve maintainability and reuse.
Collaborated with technical stakeholders to troubleshoot pipeline issues and translate requirements into reliable software improvements.

PythonImage ProcessingEmbeddingsVector RetrievalData Pipelines

Selected projects

Projects that best support software engineering roles.

Selected for implementation depth across retrieval, security, cloud systems, computer vision, and production-oriented machine learning.

01 Retrieval · LLM Application

RAG-Based Document Question Answering System

Built an end-to-end document-grounded question-answering application that converts uploaded PDFs into searchable vector representations and uses retrieved context to reduce unsupported LLM responses.

Multi-document ingestion FAISS vector search 800 / 150 chunk strategy Grounded response generation

PythonLangChainFAISSGemini APIStreamlit

View repository ↗

02 Robust ML · Computer Vision

Robust Image Classification under Spurious Correlations

Built and evaluated a Waterbirds classification pipeline to diagnose shortcut learning and improve robustness using DINOv2 representations, Grad-CAM, PCA, Core-Aligned Regularization, and Group DRO.

97.6%accuracy

94.8%F1

95.3%worst-group accuracy

PyTorchDINOv2Group DROGrad-CAMPCA

View repository ↗

03 Cloud · Security · Data

Cloud-Based Medical Data Repository

Designed a cloud-oriented healthcare platform for secure patient-data sharing and real-time ambulance tracking, combining application workflows with encryption and integrity controls.

Secure patient-data exchange Real-time ambulance tracking AES encryption SHA-256 integrity

Cloud ComputingSaaSSecurityHealthcare Systems

Additional relevant work

Earlier work that adds security and deep-learning depth to the portfolio.

Biometric Authentication in Cloud Computing

Enhanced a biometric authentication concept using AES and RSA encryption, SHA-256 hashing, and blockchain-based integrity mechanisms for cloud data protection.

PythonEncryptionCloud SecurityBlockchain

CAPTCHA Recognition System

Implemented a CNN-LSTM recognition pipeline for distorted CAPTCHA images, combining visual feature extraction with sequential decoding and achieving 99.54% accuracy.

CNNLSTMComputer VisionDeep Learning

View repository ↗

More work: Additional experiments, coursework, and earlier repositories are available on GitHub.

Browse all GitHub repositories ↗

Selected publications

Research most relevant to software and AI engineering.

Selected research spanning retrieval, NLP, computer vision, ranking, and intelligent-system development.

2024NLP · Sequence Modeling

Neural Sequence-to-Sequence Modeling with Attention for Abstractive Text Summarization

Designed a sequence-to-sequence architecture with attention to improve contextual understanding and abstractive summary generation.

View paper ↗

2024Computer Vision · OCR

Efficient CAPTCHA Image Recognition Using Convolutional Neural Networks and Long Short-Term Memory Networks

Combined CNN-based visual feature extraction with LSTM-based sequence decoding for robust CAPTCHA recognition, reporting peak accuracy of 99.54%.

View paper ↗

2024Information Retrieval · NLP

Empowering Information Retrieval: A Framework for Effective Data Summarization Using NLP and SBERT

Explored semantic retrieval and summarization using Sentence-BERT, combining morphological and semantic analysis to extract and condense information from large data sources.

View paper ↗

2024Search · Ranking · Vision

Hypergraph Fusion and Pseudo-Relevance Feedback for Enhanced Tag-Based Image Retrieval

Investigated retrieval quality using global/local visual features, hypergraph fusion, and pseudo-relevance feedback to improve tag-based image ranking.

View paper ↗

About

Software engineering, strengthened by research depth.

I am pursuing an M.S. in Computer Science and Engineering at the University at Buffalo after completing a B.Tech in Computer Science with a specialization in Cloud Computing at SRM University. I enjoy building systems that connect software engineering with machine learning: reliable data flows, retrieval, APIs and interfaces, model evaluation, cloud applications, and performance optimization.

LanguagesPython · JavaScript · SQL · C++ · C

Software & WebREST APIs · HTML · CSS · Streamlit · API Integration

AI / MLPyTorch · TensorFlow · Scikit-learn · OpenCV · NLP · RAG · LLMs

Data & RetrievalNumPy · Pandas · FAISS · LangChain

Cloud & ToolsAWS · Firebase · Git · GitHub · Jupyter · Google Colab

Education & recognition

Academic foundation.

2025 — 2027

University at Buffalo, SUNY

M.S. Computer Science and Engineering

CGPA 3.958 / 4

2021 — 2025

SRM University

B.Tech Computer Science · Cloud Computing

CGPA 9.64 / 10

#1in class in Kaggle-hosted deep-learning wildfire competition

#2 globallyin the same competition

20+student employees coordinated as a campus dining student leader

Contact

Interested in software, backend, or AI engineering work?

I’m based in Buffalo, New York and can be reached directly by email or through LinkedIn.

chakradhar3724@gmail.com LinkedIn ↗ Resume ↗