Mokshit
Surana

Jan 2025 – May 2026

Chicago, USA

Graduate Teaching Assistant @ University of Illinois Chicago

• Co-authored a research paper on LLM-generated metadata enrichment for RAG systems, accepted at IEEE CAI 2026; led experimental validation and benchmarking.
• Designed and deployed RAG infrastructure for student research teams: Graph-RAG pipelines on Neo4j, production vector stores (ChromaDB, FAISS), and clinical NLP workflows using ClinicalBERT for ICU readmission prediction.
• Mentored 9+ graduate project groups on production vector search, retrieval evaluation, and multi-agent LLM design patterns.

Feb 2023 – Jun 2024

Remote

Machine Learning Engineer @ SimPPL

• Built a scalable cross-platform data pipeline processing 80M+ comments and 500K+ videos from YouTube and Truth Social, using chunked Pandas processing and batched GPU inference under tight memory constraints.
• Engineered an automated LLM evaluation harness scoring LLaMA and Mistral 7B across 6 toxicity dimensions via the Google Perspective API, benchmarking behavior against human baselines from 460K+ messages.
• Developed a labeled taxonomy of 40 adversarial prompt patterns (jailbreak and persuasion techniques) to systematically stress-test open-source LLM safety and reproducibility.
• Conducted the research in collaboration with Prof. Deb Donig (UC Berkeley).

Jun 2023 – Jul 2023

Pune, India

Technology Intern @ Deutsche Bank

• Cut manual processing time by 96.7% by replacing a manual workflow with an automated CSV validation engine built on Spring Boot and Oracle DB.
• Built a Spring Boot PDF-generation microservice using iText, applying OOP design principles within an Agile delivery team.