AI/ML Research Engineer - NLP, LLMs, Multimodal Systems

Rohith Perumandla

Chicago, United States (open to relocate)

+1-773-***-**** ******.**********.**@*****.*** LinkedIn Portfolio SUMMARY

AI/ML Research Engineer with 4 years of experience and a strong background in Natural Language Processing (NLP) and Large Language Models (LLMs). Proven experience in designing, training, and evaluating LLM-based agents and multi-agent systems. Proficient in Python and leading AI/ML frameworks like PyTorch, TensorFlow, Hugging Face, and LangChain/LangGraph. Eager to apply my expertise in data science and high-performance computing to advance scientific discovery. EDUCATION

Depaul University Aug 2023 - Jun 2025

Masters in Data Science (GPA: 4.0/4.0) Chicago, IL,USA

• Achievements: Graduate Teaching Assistant for Advanced Machine Learning, Health in Data Science and Topics in Digital Health

• Coursework: Python ProgrammingFundamentals of Data Science, Big Data Mining, Deep Learning and Neural Networks Rajiv Gandhi University of Knowledge Technologies, IIIT Basar May 2018 - Jun 2022

• Bachelors in Electronics and Communication Engineering (GPA: 7.89/10) Hyderabad, India EXPERIENCE

DePaul University

AI/ML Research Engineer

Nov 2023 - Present

Chicago, IL

• Implemented a low-latency, multimodal conversational AI system architecture, which resulted in a 55% reduction in latency and a 60% decrease in GPU costs, significantly impacting the efficiency of real-time cognitive assessment for Dementia care.

• Built and led a cross-functional research collaboration team of 10 members, which resulted in the successful delivery of the core speech system project and 2 peer-reviewed publications, significantly impacting the AI technical development.

• Developed advanced feature selection and interpretability techniques, which improved the performance of dementia ML classification models from 60% to 76%, directly impacting the clinical utility in patient assessment.

• Designed, developed, and deployed an end-to-end AI system for dementia care on GCP, enabling scalable support for caregivers and patients through iterative development cycles incorporating real-world feedback to enhance system performance and user outcomes.

• Improved parallel processing pipelines from scratch using PySpark, to handle 200M+ physiological sensor data for Autism, which resulted in a 70% reduction in data preprocessing time, significantly impacting the stress detection in Autism participants.

• Engineered advanced text parsing and feature extraction pipelines on raw text data to design features which served as high-value features and contributed to the 16% improvement for ML/DL models. Setuserv Jan 2022 - Jul 2023

Data Scientist Hyderabad, India

• Increased NLP model accuracy by 10% (from 81% to 89%) for sentiment classification by implementing advanced text preprocessing, feature engineering, and leveraging a fine-tuned GPT-based model on large-scale e-commerce datasets.

• Generated 1 Million in new revenue through accurate demand forecasting analysis, achieved by architecting a scalable end-to-end MLOps pipeline using Airflow to process 500K+ daily e-commerce reviews using state-of-the-art NLP models (BERT, NER, Zero- shot classification) from Hugging Face.

• Delivered actionable insights to 5+ client stakeholders by performing SQL-based quantitative analysis on quarterly product data, translating complex model results into business recommendations that influenced product strategy.

• Developed 8+ ML prototypes addressing diverse business challenges by implementing rapid experimentation using Python and cloud technologies, enabling clients to validate solutions before full-scale development SKILLS

• Programming Languages: Python, R, Java, Matlab, Data Structures, SQLite

• ML/AI Frameworks: PyTorch, Keras, TensorFlow, Deep Learning, Hugging Face, Prompt engineering

• Cloud & Software Tools: Amazon Web Services (AWS), Azure, Google Cloud (GCP), Django, Git, Rest APIs

•MLOps & Tools: Docker, Git, REST APIs, MLflow, CI/CD. PROJECTS

Dental Referral Management Platform Oct 2025 – Dec 2025

• Led end-to-end design and implementation of a greenfield dental referral management platform, bridging General Practices and Specialists with secure intake, document workflows, and real-time referral lifecycle tracking.

• Designed HIPAA-aware, magic-link–based referral submission and per-patient tracking, balancing clinical privacy, low-friction adoption, and real-world dental office workflows without requiring GP account creation.

• Architected a Dentrix-compatible, read-only integration strategy for appointment and treatment status monitoring, enabling scalable care coordination while working within EHR platform constraints. AI Email Agent with LangChain, LangGraph, RAG, and MCP Integration Jun 2025- Aug 2025

• Developed and deployed an autonomous multi-agent AI email assistant using LangChain and LangGraph, powered by Retrieval-Augmented Generation (RAG) for intelligent, context-aware email response generation.

• Utilized a custom Model Context Protocol (MCP) to connect with multiple email accounts (Gmail and Outlook), maintaining memory, context, and intent history across interactions.

PUBLICATIONS

• Rohith Perumandla, Casey C. Bennett.Developing Conversational Speech Systems for Robots to Detect Speech Biomarkers of Cognition in People Living with Dementia using LLMs.(link) HOBBIES

• Building AI-powered productivity tools for macOS and iOS, including end-to-end system design and use AI to increase the productivity.

• Designing and building full software systems and MVPs using modern development tools and AI-assisted workflows(like Cursor, Antigravity, Gemini CLI).

• Staying active on LinkedIn, reading and posting blogs and research; sharing insights and engaging in discussions on LinkedIn.

Contact this candidate