CV
For downloading the PDF , click to the top right icon
Contact Information
| Name | Ekaagra Gupta |
| Professional Title | Computer science undergraduate |
| ekaagrag2006@gmail.com | |
| Location | Jaipur, Rajasthan 302026 |
Professional Summary
Pre-final year AI/ML undergrad. I build things that retrieve, reason, and explain — from satellite anomaly detectors to hybrid RAG pipelines to a Full stack Application
Experience
-
May 2026 - July 2026 Jaipur, Rajasthan
Artifical Intelligence Intern
Malaviya National Institute of Technology Jaipur (MNIT)
Engineering a production-scale Retrieval-Augmented Generation (RAG) system integrating dense retrieval, lexical retrieval, transformer-based reranking, and GPU-optimized large language model inference. The project involves memory-efficient document indexing, metadata-aware retrieval optimization, vLLM-based inference acceleration, FlashAttention, KV-cache optimization, continuous batching, and comprehensive benchmarking of latency, throughput, memory footprint, and retrieval quality across large document corpora.
-
Feb 2026 - May 2026 Lucknow , india
Artifical Intelligence and Machine Learning Intern
coreGEN
Engineered end-to-end computer vision pipelines for satellite image analysis by developing deep learning-based super-resolution models, integrating MobileNet-SSD object detection, implementing real-time anomaly detection algorithms, and deploying interactive MERN-based dashboards for visualization, inference, and model evaluation.
Education
-
2024 - 2028 Rajasthan , India
Bachelor of Technology
Manipal university jaipur
Computer Science and Engineering specalisation in Artificial Intelligence and Machine Learning
- Data Structures and Algorithms - Database Management Systems - Computer Networks - Operating Systems - Agile Software Development - Artificial Intelligence - Machine Learning - Deep Learning - Natural Language Processing - Computer Vision
-
2014 - 2024 Rajasthan , India
High School Diploma
Warren academy
Physics , chemistry and mathematics
- Physics - Chemistry - Mathematics
Awards
-
2025 Student excellence award
Manipal university jaipur
Student Excellence Award for recognition of outstanding contributions to the ISGC’25 paper and poster presentation.
-
2026 Student excellence award
Manipal university jaipur
Student Excellence Award for my recent internship at CoreGEN.
Skills
Languages
Interests
Certificates
- Prompt Engineering & Programming with OpenAI - Columbia University (2025)
- Essentials in Generative AI - Microsoft & LinkedIn (2025)
- Generative AI Studio - Google Cloud (2025)
Projects
-
MemoRAG
Engineered a memory-augmented RAG pipeline that builds a compressed KV-cache representation of long documents before retrieval, enabling globally-aware query rewriting and surrogate-question generation for multi-hop and summarization queries that stan- dard chunk-based RAG handles poorly.
- Evaluated pipeline performance on long-context QA and summarization tasks, comparing standard RAG against the memory-augmented approach for retrieval accuracy and answer completeness
- Integrated a dense retrieval layer using FAISS and BGE-M3 embeddings, paired with a lightweight variant (Qwen2.5-1.5B) for lower-resource deployment.
-
GETHER
Developed a hybrid spatio-temporal air quality forecasting framework combining stacked LSTM networks, causal discovery, and explainable AI to model pollutant interactions and predict Air Quality Index (AQI). The pipeline integrates temporal feature engineering, Granger causality analysis, SHAP-based interpretability, and counterfactual policy simulation for transparent environmental decision support.
- Built a stacked LSTM forecasting model with engineered lag and rolling-statistic features, benchmarked against Linear Regression and Random Forest baselines, while optimizing hyperparameters for improved forecasting accuracy.
- Designed a causal inference and explainability module using Granger causality, dynamic causal graphs, SHAP feature attribution, uncertainty estimation, and counterfactual simulations to evaluate the impact of emission-control policies on future AQI.
-
SEVAS
Engineered an AI-powered predictive environmental monitoring platform for illegal sand mining and land encroachment detection using multi-temporal satellite imagery, deep learning, and geospatial analysis. The system combines computer vision, spectral analysis, and temporal forecasting to enable proactive environmental enforcement.
- Developed a hybrid vision pipeline integrating a custom U-Net segmentation network, Gemini Vision, NDVI/NDWI spectral indices, and temporal LSTM forecasting to detect environmental degradation, localize mining regions, and predict future violations.
- Built a production-ready MERN application with RESTful APIs, automated severity scoring, cross-jurisdictional intelligence, and interactive geospatial visualizations for large-scale environmental monitoring and decision support.