cv

Advanced Data Science Associate

ZS

Building GenAI based healthcare solutions for pharma.

• Building GenAI based healthcare solutions for pharma
• Technologies: GenAI, healthcare AI workflows, Python

Predoc Associate

DRDO & Indian Institute of Science, Bangalore

Worked on ML systems for combat aircraft applications and SciML research at AiREX lab under Prof. Sashikumaar Ganesan.

• Developed low level C++ components to integrate ML models for CASDIC's combat aircraft hardware for low-latency inference
• Evaluated fine-tuned IBM Granite & Gemma models for SciML using a custom Graph-RAG pipeline, achieving SOTA accuracy on benchmark datasets
• Led open source projects PULSE & SciREX at AiREX under Professor Sashikumaar Ganesan
• Technologies: C++, IBM Granite, Gemma, Graph-RAG, SciML

Data Scientist

Bajaj Finserv Health

Worked on fraud detection models and graph-based pipelines.

• Designed and implemented a high-performance statistical model for fraudulent transaction prediction, achieving the highest abuse detection rate
• Spearheaded end-to-end development of a fraud detection pipeline using graph database technologies (Memgraph, Neo4j, Cypher)
• Built and optimized SQL queries to aggregate features from relational databases, improving training data quality for fraud detection models
• Technologies: Python, SQL, Memgraph, Neo4j, Cypher

Machine Learning Intern

IIT Indore

Developed ML models for financial forecasting.

• Developed ML models for stock-price prediction using Python, achieving 85% accuracy by incorporating news and market data
• Implemented time-series analysis and feature engineering, boosting model accuracy by 25%
• Technologies: Python, time-series modeling, feature engineering

Open Source Technical Writer

OpenMined

Contributed to documentation and technical blogs for privacy-preserving ML.

• Collaborated with the writing team to improve existing documentation and beginner tutorials in differential privacy
• Published blog posts on federated learning and privacy-preserving machine learning, resulting in increased product adoption
• Technologies: Differential privacy, federated learning, technical writing

B.Tech

National Institute of Technology, Hamirpur

Electrical Engineering

Disentangling Recall and Reasoning in Transformer Models through Layer-wise Attention and Activation Analysis

AAAI XAI4Science Workshop 2026

Can Linguistically Related Languages Guide LLM Translation in Low-Resource Settings?

LoResMT 2026: EACL 2026 Workshop

The Evolution of FlashAttention

ICLR Blog Post Track 2026

RADAR: Mechanistic Pathways for Detecting Data Contamination in LLM Evaluation

NeurIPS 2025 LLM Evaluation Workshop

Equilibrium Dynamics and Mitigation of Gender Bias in Synthetically Generated Data Evaluation

AAAI Workshop on Shaping Responsible Synthetic Data in the Era of Foundation Models

AIFred – Secure, Local AI Coding Assistant

Desktop AI coding assistant with real-time LLM and ASR, ensuring 100% local data privacy.

• Launched AIFred: a desktop coding assistant with real-time AI and 100% data privacy via a local AI pipeline (Ollama for LLM and NVIDIA NeMo for ASR)
• Built an entirely on-device voice command interface for secure offline querying
• Optimized for low latency by architecting a fully local AI pipeline and shipped an Electron overlay UI for seamless coding workflows
• Technologies: Python (FastAPI, NeMo, pydub), JavaScript (Electron), Ollama API

Document Tampering Detection using YOLOv8

Forgery detection system using YOLOv8 and Streamlit.

• Developed a forgery-detection system with YOLOv8 to identify overwriting, scribbling, and whitener use
• Built a dataset of 600+ document images for robust detection
• Designed a Streamlit interface for document upload, tampering visualization, and detailed digital-forgery analysis
• Technologies: Python, Streamlit, OpenCV, YOLOv8

Fundus AI

AI-powered ophthalmic disease detection solution.

• Developed an end-to-end solution to detect ophthalmic and systemic diseases such as Diabetic & Hypertensive Retinopathy, Cataract, Glaucoma, and ARMD
• Built a deep-learning smartphone app for accessible ophthalmic screening
• Designed a 3D-printed Tinkercad-based accessory for high-quality fundus image capture
• Technologies: Python, Keras, TensorFlow, Flutter, Tinkercad

Teaching Assistant

IISc Bangalore

DS-246, Generative & Agentic AI in Practice. Delivered lectures and tutorials, led doubt sessions, and was responsible for developing and grading assessments.

Machine Learning & Data Science Trainer

DRDO Industry Academia Centres of Excellence (DIA-CoEs)

Designed & delivered a 3-day training module to DRDO scientists, focusing on foundational Machine Learning and core data science tools.