I’m good at finding patterns, connecting ideas, and looking at problems from different perspectives. I enjoy figuring out how things work beneath the surface rather than just accepting what I see. I’m naturally curious, persistent when faced with challenges, and always looking for smarter ways to turn complex problems into simple solutions.
Associated Data Scientist
Soulax Software Pvt. Ltd.
- Designed and implemented production-grade multi-tenant Django REST APIs for lien search, summaries, waivers, and payouts, with secure PostgreSQL schema-per-tenant isolation via django-tenants.
- Built LLM-based document pipelines from OCR to structured JSON and summarization using AWS S3 inputs.
- Optimized LLM batching and budget strategy (Gemini 2.5 Pro vs 3 Pro), achieving up to 96% accuracy with 70% lower cost and 61% faster processing.
- Owned end-to-end backend and AI pipeline development from design to deployment, including retry logic, monitoring, and traceable logging.
- Managed Jira-driven sprint delivery, maintained GitHub branch and PR workflows, and supported Dockerized services with Jenkins-based CI pipelines.
AI/ML Intern
Soulax Software Pvt. Ltd.
- Developed an automated LLM-based summarization pipeline for fixed-format PDFs, generating structured DOCX outputs for operational reporting.
- Benchmarked OCR and LLM combinations for VIN extraction from embedded PDF images, achieving 98% accuracy with Gemini 2.5 Flash.
AI/ML Intern
Shorter Loop
- Built a fault-tolerant AWS document ingestion pipeline with multithreading and retry logic, maintaining 95% uptime for a RAG workflow.
- Engineered Sentence-BERT + NLTK retrieval pipelines, indexing 100K+ vectors in Qdrant for scalable semantic search.
Frequency-Domain Image Watermarking (Diffusion Models)
Built a DCT-domain watermarking system integrated with Stable Diffusion, balancing watermark strength with image fidelity in generated outputs. Validated robustness against scaling, rotation, compression, and frequency-domain attacks, achieving >90% extraction accuracy under transformations. Reported strong evaluation metrics with AUC 0.95, TPR@1%FPR 0.92, FID 30, and CLIP Score 33 across multiple watermark scales.
Web Page Classification with Machine Learning
Built an NLP classification pipeline over 100K+ web page samples using tokenization, lemmatization, and TF-IDF feature engineering. Trained and evaluated Naive Bayes and Logistic Regression models, achieving 95% classification accuracy on validation datasets.
M.Tech., Computer Science
Sri Sathya Sai Institute Of Higher Learning · 8.1
M.Sc., Mathematics (Specialization in Computer Science)
Sri Sathya Sai Institute Of Higher Learning · 7.8
B.Sc., Mathematics
Sri Sathya Sai Institute Of Higher Learning · 6.5
GATE 2023 (Computer Science)
AICTE
Mar 2023AWS Certified AI Practitioner
Amazon Web Services
Oct 2025Amazon GameDay Hackathon
Amazon Web Services
Feb 2026Soulax Hackathon
Soulax Software Pvt. Ltd
May 2026
Attendee · Oct 24, 2026