University of California San Diego
Sep 2025 - PresentHalıcıoğlu Data Science Institute
M.S. in Data Science
shashank@257:~$ whoami
MS Data Science @ UC San Diego
Researcher who publishes. Engineer who ships. Founder who builds movements.
I architect production-grade, end-to-end AI systems at the frontier of generative AI and LLMs, from LLM-powered RAG pipelines and transformer architectures to adversarially-robust deep learning, shipping production AI/ML systems end to end. A globally-ranked hackathon champion and 0→1 founder, I turn state-of-the-art research into products people actually use, and I move fast.
Halıcıoğlu Data Science Institute
M.S. in Data Science
Jawaharlal Nehru New College of Engineering
B.E. in Computer Science & Engineering
Where others see busywork, I see automation. Supporting the administrative operations of Graduate Family Housing, from customer service to data entry, I take the initiative to automate manual, repetitive workflows and build internal tools that save staff time.
Spearheaded end-to-end ML pipelines at enterprise scale: data processing, cleaning, EDA, feature engineering & NLP across 100k+ business records, shipping production predictive & time-series forecasting models delivering an elite 0.98 F1 score. Partnered across teams to interpret and optimize model outputs for data-driven decision-making. (Digitide, formerly Conneqt Digital.)
Owned product strategy, roadmap & workflow architecture for a university-backed AgriTech e-commerce marketplace, built for startups to onboard and sell, targeting real agricultural supply-chain gaps with technology-driven, data-informed design.
Built the data & analytics backbone of WaitKyu, a queue-management platform serving 2,000+ active users, shipping ML-powered insights and rigorously tested features that measurably elevated UX and product decisions.
Designed, shipped and maintained production full-stack web applications end-to-end, owning everything from UI to backend integration.
Developed predictive machine learning models for classification & survival analysis, reaching ~92% accuracy through rigorous feature engineering, exploratory data analysis, and ML pipeline optimization.
Built applied machine learning workflows spanning model training, evaluation, and real-world dataset analysis, with a focus on performance validation and model robustness.
Completed structured cybersecurity training in threat analysis, vulnerability assessment, and core security frameworks applied to real-world environments.
Engineered robust, high-performance backend systems in Python & Java while mentoring junior engineers, driving debugging, feature development, and system-level improvements across live development workflows.
Founded and scaled a technology & entrepreneurship movement from zero to a 5,000+ member community within months, importing industry-level structure into a university that had almost none. Delivered founder-led sessions, technical workshops, hackathons & startup networking, secured national-level hackathon partnerships, and earned a seat in institutional decisions that reshaped student life. Now mentoring the next generation of leaders.
Led a city chapter of India's largest open-source community (20,000+ members), driving national conferences, city chapters, hackathons & technical events, championing open-source software nationwide and connecting students with industry at scale.
Rose from volunteer to national lead for Parrot Security OS, a globally-used Linux distribution with 50M+ downloads, going on to help manage and moderate its 95.5K+ member global community. Drove operations, open-source contributions & user engagement.
Co-founded and scaled an international cybersecurity community to a 10,000+ member platform, advancing ethical pen-testing, hands-on skill development & collaborative security research across a global member base.
Flagship deepfake audio detection system. SVM pipeline engineered on MFCCs, chroma, spectral centroid, bandwidth & ZCR, achieving 95.7% validation accuracy on large-scale datasets. Peer-reviewed, published and secured research funding. Advances AI-driven media forensics, cybersecurity & content integrity.
Pioneered a deep learning CNN pipeline detecting colorectal cancer from vocal biomarkers: a non-invasive, low-cost, AI-powered screening breakthrough with real clinical decision-support potential.
Novel hierarchical deep learning + ML intrusion detection architecture with a purpose-built adversarial-example detector, delivering measurable accuracy gains under attack while optimizing compute for evolving threat landscapes.
Comprehensive threat-model taxonomy classifying attacks and mapping them to defenses across major deep learning frameworks. A practical blueprint for Responsible & Secure AI in production.
Product-led AgriTech e-commerce platform built with a university agri-business incubator, connecting farmers and startups to tackle real supply-chain challenges with technology-driven, data-informed solutions.
Reproducibility study of LipsLev (ICLR 2025) on certified robustness of text classifiers under bounded Levenshtein distance: worked through the math, surfaced implementation bugs, and ran experiments on bound tightness. Published in the Trustworthy Machine Learning publication.
🏆 1st Place & #6 Globally at the SmashHack ML Hackathon (UC San Diego & NSF HDR), building a climate-tech machine learning system under time pressure, ranked against a worldwide field. post ↗
🎖 Outstanding Achievement & Contribution Award from Jawaharlal Nehru New College of Engineering, for academic excellence and standout leadership across clubs, hackathons & technical communities.
🥈 Runner-Up at the Hack for Hire Hackathon (Anvesana TBI), for an AgriTech product solving real-world agricultural challenges. post ↗
📄 4× Peer-Reviewed Publications spanning AI, cybersecurity & healthcare: deepfake detection, medical deep learning, adversarial robustness & secure AI, with research funding secured.